THE INFLUENCE OF INFORMATION AND PRODUCT LABELS ON CONSUMER

THE INFLUENCE OF INFORMATION AND PRODUCT LABELS ON CONSUMER

PREFERENCES AND WILLINGNESS TO PAY FOR PECANS

by

SHARON P. KANE

(Under the Direction of Gregory Colson)

ABSTRACT

Pecans, with a $560 million U.S. agricultural production value, remain a largely

unexamined area of consumer preferences, willingness to pay (WTP) for attributes, or valuations

of mandatory country of origin (COO) labeling requirements. Recent approval of a Federal

Marketing Order (FMO) for the U.S. pecan industry highlights the importance of consumer

valuation as stakeholders pursue avenues to address marketing and consistency-of-standards

challenges in this sector. Key to investigating consumer valuations are product labels, on which

vital information is conveyed by producers and manufacturers to consumers at point of sale.

Clear and effective labels not only allow consumers to align their preferences with available

options, but hold potential economic gains when product information garners a price premium

for desired attributes.

Employing choice experiments and random nth price experimental auctions in a series of

research sessions with adult consumers in two Southeastern U.S. cities, this study assesses how

consumers respond to product attributes and label details in different information contexts for

purchases of shelled pecan halves. From the data collected in the research sessions, we address

the following: 1) consumers’ willingness to pay for select initiatives proposed in conjunction

with the FMO regarding pecan attributes, 2) directly eliciting the consumers’ value of obtaining

information about attributes vs. WTP for particular attributes in pecan purchases under

alternative labeling scenarios, and 3) the influence of consumer risk and ambiguity attitudes on

consumer preferences for single or mixed country of origin products.

Findings indicate that, in choice experiments, consumer ethnocentric tendencies and

purchasing patterns play a role in defining differences among consumer taste preferences and

WTP, while efforts to educate consumers about attributes are essential. Overall, despite

significant taste heterogeneity, consumers generally are most willing to pay a premium for

pecans of U.S. origin over other attributes, though there is some evidence from the experimental

auctions that the value of origin information diminishes in a more complex information

environment. However, in the presence of risk or ambiguity regarding knowledge of product

provenance, consumers prefer mixed origin over any risk of obtaining the single origin product

from their lowest ranked preference country.

INDEX WORDS: value of information, willingness to pay, experimental auctions, consumer

preferences for pecans



by

SHARON P. KANE

BS, University of Tennessee at Chattanooga, 1986

MS, University of Kentucky, 1998

A Dissertation Submitted to the Graduate Faculty of The University of Georgia in Partial

Fulfillment of the Requirements for the Degree

DOCTOR OF PHILOSOPHY

ATHENS, GEORGIA

2018

© 2018

Sharon P. Kane

All Rights Reserved



by

SHARON P. KANE

Major Professor: Gregory Colson Committee: John Bergstrom Benjamin Campbell Electronic Version Approved: Suzanne Barbour Dean of the Graduate School The University of Georgia May 2018

iv

ACKNOWLEDGEMENTS

This work is supported by SCRI Award No. 2011-51181-30674 and Award No. 2013-

51106-21234 from the USDA National Institute of Food and Agriculture. Any opinions,

findings, conclusions, or recommendations expressed in this publication are those of the

author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture.

First, I would like to express my sincere gratitude to Dr. Greg Colson for his economic

expertise, honest input, and unflappable patience that propelled this idea into reality. His

guidance and support in research design and writing, along with thoughtful consideration for all

of my questions was invaluable. I would also like to thank the other members of my committee

for the generosity of their time and ideas in contributing to this project: Dr. John Bergstrom for

assistance, encouragement and feedback from the beginning of this quest, contributing his

knowledge in topic discussions and helping navigate the process, Dr. Ben Campbell for lending

his particular expertise in latent class analysis, brainstorming of ideas, and for asking the hard

questions that matter. I would like to thank Dr. Kent Wolfe for championing these efforts from

inception, along with the rest of the Center for Agribusiness and Economic Development, whose

patience and assistance never failed. I am grateful for Dr. Jim Bason’s aid in making the

research sessions run smoothly, Dr. Ron Pegg for securing the pecans needed in conducting this

research, Dr. Koushik Adhikari and sensory laboratory staff for organizing and recruiting

participants, Grace Melo for packing pecans into bags, and Soye Shin for assistance during

experiments. Last but not least, thanks to my husband Patrick, my parents, sisters, children and

grandchildren, all of whom inspired and supported me along the way.

v

TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS ........................................................................................................... iv

LIST OF TABLES ........................................................................................................................ vii

LIST OF FIGURES ....................................................................................................................... ix

CHAPTER

1 AN EX-ANTE ASSESSMENT OF CONSUMER RESPONSE TO PROPOSED

PECAN INDUSTRY FEDERAL MARKETING ORDER INITIATIVES ..................1

Introduction ..............................................................................................................1

Research Design/Data Collection ............................................................................4

Research Method: Mixed Logit, Latent Class Logit, and WTP ...........................10

Findings and Discussion ........................................................................................14

Conclusion .............................................................................................................20

References .............................................................................................................24

2 USING EXPERIMENTAL AUCTIONS TO DIRECTLY MEASURE THE VALUE

OF INFORMATION IN PECAN PURCHASES ........................................................28

Introduction ............................................................................................................28

Methods and Procedures ........................................................................................34

Data Summary and Results ....................................................................................44

Conclusion .............................................................................................................50

References ..............................................................................................................56

vi

3 CONSUMER PREFERENCES FOR SINGLE VS. MIXED COUNTRY OF

ORIGIN: EVIDENCE FROM EXPERIMENTAL AUCTIONS ON THE

INFLUENCE OF RISK AND AMBIGUITY ATTITUDES.......................................62

Introduction ............................................................................................................62

Methods and Procedures ........................................................................................66

Data Analysis and Results .....................................................................................79

Conclusion .............................................................................................................84

References ..............................................................................................................88

4 GENERAL CONCLUSION ........................................................................................92

APPENDICES

A PARTICIPANT INFORMATION PACKET ..............................................................95

vii

LIST OF TABLES

Page

Table 1.1: Choice experiment category descriptions .......................................................................6

Table 1.2: Sample descriptive statistics ...........................................................................................9

Table 1.3: Consumer label-reading frequency and organic purchasing ........................................10

Table 1.4: Mixed logit model parameter estimates and consumer WTP .......................................17

Table 1.5: Latent class model parameter estimates .......................................................................22

Table 1.6: Latent class model estimates of consumer WTP ..........................................................23

Table 2.1: Summary statistics ........................................................................................................35

Table 2.2: Container label options for auction ...............................................................................42

Table 2.3: Summary statistics of auction bid prices ......................................................................46

Table 2.4: Mean bid difference between pairs of label treatments ................................................47

Table 2.5: Mean bid difference between pairs of label treatments for positive bids .....................48

Table 2.6: OLS estimates: All bids ................................................................................................51

Table 2.7: Tobit estimates: All bids ...............................................................................................52

Table 2.8: OLS estimates: All bids except for all-zero across treatments .....................................53

Table 2.9: Tobit estimates: All bids except for all-zero across treatments ....................................54

Table 2.10: OLS estimates: Bid differences ..................................................................................55

Table 3.1: Summary statistics ........................................................................................................67

Table 3.2: Label treatment options ................................................................................................75

Table 3.3: Aggregate risk and ambiguity attitudes under real and hypothetical incentives ..........79

viii

Table 3.4: Summary statistics of auction bid prices ......................................................................80

Table 3.5: Mean bid difference between all pairs of label treatments ...........................................81

Table 3.6: Relevant bid differences and definitions ......................................................................83

Table 3.7: Regression estimates for bids .......................................................................................86

Table 3.8: Regression estimates for bid differences ......................................................................87

ix

LIST OF FIGURES

Page

Figure 1.1: Example of choice set question .....................................................................................7

Figure 2.1: Details of experiment steps .........................................................................................38

Figure 2.2: Example of practice round direct VOI experiment task ..............................................40

Figure 2.3: Example of auction label treatment round bid sheet ...................................................43

Figure 3.1: Details of Experiment Steps ........................................................................................71

Figure 3.2: Example of practice round for auction ........................................................................73

Figure 3.3: Example of Auction Label Treatment Round Bid Sheet .............................................76

1

CHAPTER 1: AN EX-ANTE ASSESSMENT OF CONSUMER RESPONSE TO PROPOSED

PECAN INDUSTRY FEDERAL MARKETING ORDER INITIATIVES

Introduction

The United States is the world’s largest producer of pecans, a native North American tree

nut (Wood, Payne, and Grauke 1990, Wells 2017) that represents a production value of over

$560 million, with approximately 80 percent of the utilized pecan production and dollar value

originating from three states – led by Georgia, followed by New Mexico and Texas (USDA

2016b). Pecans comprise about four percent of U.S. tree nut production (in-shell basis) and just

under eight percent of production value, in a category for which almonds command more than

two-thirds of the $7.7 billion total tree nut production value (USDA 2016a). According to the

United States Department of Agriculture Outlook, U.S. pecan production is expected to increase

three percent to over 262 million pounds for the 2016/17 marketing season (USDA 2017).

In 2016, a Federal Marketing Order (FMO) for the 15-state pecan production region1 of

the U.S. was approved and codified into regulation with the goal of benefitting those who grow,

buy, process, and eat pecans (American Pecan Board 2016). Initiated as industry-driven

agreements, FMOs are individually tailored to promote an industry and cooperatively address its

customized needs, ultimately becoming binding regulation after approval by producers and the

Secretary of Agriculture (USDA). The 2016 FMO was obtained by the pecan industry citing

1 The FMO covers pecans grown in Alabama, Arizona, Arkansas, California, Florida, Georgia, Kansas, Louisiana,

Mississippi, Missouri, New Mexico, North Carolina, Oklahoma, South Carolina, and Texas. Any financing required

to carry out directives of the FMO are handled through assessments on the handlers of pecans grown in these states

and locally managed, under USDA oversight, by the American Pecan Council which is assembled as a result of the

order. (USDA 2016) For more detail, see the final rule, effective August 5, 2016 (Federal Register 2016).

2

challenges that include supply concerns, prices, market disruption, and lack of uniform quality,

container, and packaging standards (American Pecan Board 2017). The final rule encompasses

several actions to be undertaken to address these issues, including collection of industry data on

production, inventory, and supply, authorization of funding for research on health and nutrition

aspects of pecans and improved technology, as well as recommendations for uniform grade, size,

quality, and container standards (Federal Register 2016).

Within the final ruling, there is documentation of the economic benefits to the growers

and handlers, based on generic results that agricultural product promotion stimulates demand and

ultimately translates into higher prices for producers (Federal Register 2016). However, within

the literature there is no direct evidence demonstrating just how consumers might respond to

potential actions prescribed by the pecan FMO being implemented in the marketplace,

particularly with respect to grade, size and quality standards or health and nutrition

characteristics.

In general, pecans are largely an understudied commodity in terms of consumer

preferences. The primary emphasis of available research includes a focus on native versus

improved pecan varieties (Palma, Collart, and Chammoun 2015), pecan consumer demographics,

consumption frequency, purchase patterns and overall tree nut nutrition knowledge (Lillywhite,

Simonsen, and Heerema 2014, Lombardini, Waliczek, and Zajicek 2008), festival attendees

preferences among three tree nuts (Gold, Cernusca, and Godsey 2004), quality perception (Park

and Florkowski 1999) and factors affecting retail outlet selection in purchasing pecans

(Florkowski, You, and Huang 1999).

Further unexplored is how consumers might respond to country-of-origin (COO) labeling

policies related to pecans. The mandatory point-of-purchase COO labeling for beef and pork has

3

recently been discontinued in the U.S., yet the requirement for pecans and other “covered

commodities” remains in place as a regulatory policy with the aim of better-informing

consumers (USDA). True to the stated intent of the regulation, some studies have found that

COO labeling may help to mitigate the search costs for consumers who prefer domestic food

products and functions as a solution to asymmetric information (Lusk et al. 2006). However,

geographic information may also be perceived as a quality or food safety marker (Lim et al.

2013, Lewis and Grebitus 2016), a way to connect with or support one’s region or locality (Boys

and Blank 2016) reflect consumer ethnocentrism (Lusk et al. 2006, Lewis and Grebitus 2016,

Klain et al. 2014), represent a freedom-of-speech issue (Tushnet 2015) or provide other signals

about the product (Lusk and Briggeman 2009). Though the consumer decision-making process

can be complex (Lusk et al. 2006, Deselnicu et al. 2013, Costanigro et al. 2014) and most studies

are context-or product-specific, there is limited evidence that consumers generally favor labeling

identifying the country of origin (Lim et al. 2013), but any such substantiation for pecan

purchases requires further exploration.

In order to evaluate how pecan consumers might respond to potential FMO

recommendations and COO labeling, we examine adult consumer preferences for shelled pecan

halves with organic, geographic origin, freshness, grade, size and health/nutrition attributes via

in-person choice experiments in two Southeastern U.S. cities. The product characteristics

selected for the choice experiments were modeled after the FMO framework in order to address

current and future challenges for the pecan industry and fill the gap in the literature with

improved knowledge of consumer preferences for pecans.

In the remainder of the paper, we detail the choice experiment design, experimental

procedures, and sampling strategy. Following key sample statistics, estimates of consumer

4

preferences and willingness to pay for pecan attributes are presented employing two alternative

econometric specifications - mixed logit and latent class models. Finally, we conclude with a

discussion of the implications of the results for the potential of the FMO recommendations to

increase pecan demand.

Research Design/Data Collection

Recruitment

Consumer recruitment took place in two communities in Georgia—Griffin, and Athens—with

the research sessions occurring at University of Georgia research facilities. For the Griffin

sessions, consumers were recruited in the summer of 2016 from the Sensory Evaluation and

Consumer Lab (University of Georgia Griffin campus) database of adults in the general public,

with 240 agreeing to participate at the time of recruitment. Participants were recruited and pre-

screened for tree nut allergies2 using an online survey for the one-hour sessions. Following the

screener check, participants were contacted and chose from offered sessions, with several being

held over a three-day period at different time slots with a target of approximately twenty

participants per session. Each person was reminded once, by either email or phone, the day

before the scheduled research would take place. A total of 218 participants showed up for

sessions held on the campus in Griffin.

A flyer with recruitment information was posted in locations throughout Athens,

containing an email to contact for additional information and scheduling. Upon receipt of the

email, we sent additional information to responders via an online survey, which included pre-

screening for tree nut allergies and selection of an offered time slot for research sessions over a

three-day period. Each person was reminded once by email the day before the scheduled research

2 Though there would be no consumption of pecan products on-site, research was restricted only to those without nut

allergies in an abundance of caution for the well-being of participants and this method was how the IRB approved

the study.

5

would take place. A total of 75 people participated in the research sessions in Athens. In both

locations, each group of interested respondents was informed that they would take part in

research about consumer preferences for tree nuts and given a participation fee of $50 upon

arrival for a research session. Each participant was given a questionnaire that included basic

demographics (gender, education, age, race, and income), pecan consumption and buying habits,

awareness of mandatory country of original laws with respect to pecans, label reading behavior

with common purchases, and questions that allow a calculation of their ethnocentrism as outlined

in Klain et al. (2014).3 The ethnocentrism measure—a 1 to 5 scale with higher indicating more—

addresses the contention that differences in consumer valuation by country of origin are driven

merely by consumer protectionism preferences. Across the 15 sessions held over a two-week

period, 293 participants completed the study, close to meeting the goal of 300 respondents.

Choice Experiment Design

Under the objective of illuminating consumer preferences and willingness to pay within the

constructs of potential new standards, exact combinations of these product attributes do not

currently exist in the marketplace. Therefore, we create hypothetical combinations to question

respondents and maintain ethical standards of research design (Colson et al. 2016).

The choice experiment portion of the study uses a balanced orthogonal design with

NGENE software producing 2 blocks of 8 choice sets each. Each choice set offers the

respondent two selections from which to purchase an 8 oz. bag of shelled pecan halves with

different attributes and prices (see Figure 1.1 for an example of the choice set question, similar to

3 We used the same method as Klain et al. (2014), which involved shortening the CETSCALE found in Shimp and

Sharma (1987) from a set of 17 Likert-scaled question down to three questions. Their test of the original measure led

to choosing the three that had the highest factor loadings in the general ethnocentrism scale, leading to development

of a 3-question based average. This average was calculated based on the respondent’s answer to three questions

where each individuals’ responses ranged from “strongly agree” to “strongly disagree” on a five-point Likert-scale.

The exact language in the three questions as quoted from Klain were: “Americans should not buy foreign products,

because this hurts American business and causes unemployment,” “It is not right to purchase foreign products

because it puts Americans out of jobs,” and “A real American should always buy American made products.”

6

Lusk, Roosen, and Fox 2003). Another selection is to choose not to purchase either option

resulting in three alternatives within each of the 8 sets. Following (Lim et al. 2013) and others,

one of the three choices is the “would-not-buy” option. Unlike some product purchases, for a

bag of shelled pecans, this is a reasonable alternative with little possibility for misinterpretation

(McFadden 2015) given likely familiarity with pecans among participants, making for a more

realistic purchasing environment in which the respondent can choose neither option (Hensher

2010).

The selections within the choice sets consist of seven overall categories: price, organic

production, country of origin, expiration date, size, grade, and health or nutrition claim, with

more detailed choices within the expiration, size, grade, and health or nutrition categories. A

summary of the attributes and levels are presented in Table 1.1. The approximate retail price of

an 8-ounce bag of shelled pecans (not organically produced) at a retail outlet was $5.99 at the

time the data was collected.4 Other categories include country of origin selection (The United

States or Mexico), whether the product was organically produced, and how far in the future the

product would expire (3, 6, or 12 months). All other attributes in the choice sets are based on

FMO objectives regarding size, grade, and health or nutrition claims as outlined in the

introduction of this report.

Table 1.1: Choice experiment category descriptions

Pecan Attributes Pecan Levels

Price $2.25, $5.00, $8.50

Organic No, Yes

Expiration Date 3 months, 6 months, 12 months

Country of Origin Mexico, United States

Size None indicated, Small, Large, Extra Large, Jumbo, Mammoth

Grade None indicated, Standard, Choice, Fancy

Health/Nutrient Claim None indicated, Heart Health, Naturally High in Antioxidants

4 Research sessions took place in July and August of 2016. Organic shelled pecans in 8 oz. bags are not commonly

found in area brick-and-mortar grocery stores. For perspective on the size of the organic pecan market, USDA 2014

Organic Survey reports only 1.5% of total pecan production in 2014 was organic.

7

Figure 1.1: Example of choice set question

Summary Statistics

Basic summary statistics of participant sample are shown in Table 1.2. Nearly two-thirds of the

sample are female (61.8%) and most participants are between 18 and 34 years of age, but

participant ages range from 18 to 77. Marital status, racial/ethnic distribution, and income and

education levels closely reflect those from the region in which they live. Participants are asked

about how often they consume pecans, ranging from very rarely to very often, with the largest

response being “sometimes” at 45%. Just over one-third (34%) answered either often or very

often, with the Griffin research participants more often than those at the Athens site (38% vs.

23%).

Consumers are also queried regarding their awareness of mandatory country-of-origin

(COO) labeling for pecans and their purchasing patterns. Only 22.6% indicate that they know

COO labeling exists for pecans. As defined in Klain et al. (2014), we average responses to

generate each respondents’ overall ethnocentrism score. Unlike the Klain findings, our

participants are, on average, less ethnocentric than not with an average score of 2.7 (2.4 in

Athens, 2.8 in Griffin), slightly less than the midpoint of the size point scale.

8

Label Information

The questionnaire provides further details about the consumers’ characteristics related to

purchasing preferences (Table 1.3.). About 63% indicate that they either always (26.4%) or

often (37.0%) read food package labels. More than 4 out of 5 say that they either always

(57.9%) or often (24.3%) look for expiration dates, while less than one-half of that amount

(32.9%) report always or often looking for the country of origin on labels. Slightly less than

twenty percent purchase organic foods either always (6.9%) or often (12%).

Consumer-reported Diet and Health

Respondents are asked to rate the healthiness of their diet and their overall health on a 1 to 10

scale, with 1 being the least healthy and 10 the most. Respondents tend more towards believing

both their diet and their overall health are more healthy than unhealthy, with the median response

for both questions at seven. The mean response for diet is 6.4 and health at 7.0.

9

Table 1.2: Sample descriptive statistics (n=289)

Variable Description Sample

Percent/Mean

Gender Female 61.8%

Age 18-34 years 42.9%

35-44 years 15.2%

45-54 years 21.5%

55-64 years 18.0%

65 years + 2.4%

Marital Status Married 47.1%

Single 41.9%

Divorced/Widowed 8.3%

Other 2.8%

Race African-American 13.8%

Asian-American 5.5%

Caucasian 72.2%

Hispanic/Latino 2.4%

Other 6.2%

Income Less than $25,000 25.5%

$25,000 to $99,999 58.3%

More than $100,000 16.2%

Education Bachelor’s or higher 37.6%

Eat pecans Very Often 10.0%

Often 24.4%

Sometimes 45.0%

Rarely 15.5%

Very Rarely 5.2%

Knowledge of COO Yes 22.6%

No 55.8%

Don’t Know/Not sure 21.6%

Ethnocentrism 1=low, 5 = high (mean) 2.7

10

Table 1.3: Consumer label-reading frequency and organic purchasing

Percent of Participants

Always Often Sometimes Rarely Never

Read Nutrition Labels 26.4% 37.0% 28.1% 5.1% 3.4%

Look for Expiration Dates on Labels 57.9% 24.3% 13.4% 3.1% 1.4%

Look for Country of Origin on Labels 12.3% 20.6% 31.2% 25.3% 10.6%

Purchase Organic Foods 6.9% 12.0% 34.7% 28.2% 18.2%

Research Method: Mixed Logit, Latent Class Logit, and WTP

Given their recognition as the essential method to study discrete choice data, we utilize the logit

family of models (Hensher and Greene 2003). Specifically, because consumers are likely

heterogeneous across tastes and preferences (Wedel and Kamakura 2000), we employ the mixed

logit and latent class logit models in order to capture any potential unobserved differences

(Wedel and Kamakura 2000, Boxall and Adamowicz 2002, Greene and Hensher 2003, Kafle,

Swallow, and Smith 2015). Each model possesses specific merits for applicability (Greene and

Hensher 2003).

Random utility theory (McFadden 1974) provides the economic framework for the

empirical analysis from the perspective of consumer utility maximization under both methods.

Within this context, we begin with the utility function (𝑈𝑖𝑗𝑡 ) of consumer 𝑖 for alternative 𝑗 in

choice set 𝑡:

𝑈𝑖𝑗𝑡 = 𝐱𝑖𝑗𝑡𝜷 + 𝜀𝑖𝑗𝑡,

where 𝐱𝑖𝑗𝑡 is a vector of the attributes of alternative 𝑗 in choice 𝑡, 𝜀𝑖𝑗𝑡 is the error term, and 𝜷 is

to be estimated. Under the assumption of a utility maximizing agent, with each choice set 𝑡 the

consumer chooses the alternative yielding the greatest utility. As demonstrated by McFadden

(1974), with the error term independent and identically distributed (iid) extreme value (Hole

11

2007), the resulting choice probability in the logit context is denoted by consumer 𝑖 choosing

alternative 𝑗 in choice situation 𝑡:

𝑃𝑖𝑗𝑡 = exp (𝐱𝑖𝑗𝑡𝜷)

∑ exp(𝐱𝑖𝑘𝑡

𝜷)𝐽𝑘=1

.

Mixed Logit Model

Building upon this framework, the mixed logit model extends the basic Logit model to allow the

estimated parameters, 𝜷, to be random instead of fixed, which enables heterogeneous taste

preferences of the consumers to be incorporated (Hole 2007b). The estimated random 𝜷 has a

distribution of 𝜷~h(𝜽 + 𝐯,) which allows researchers the flexibility to apply to the random

parameters h(.) any appropriate probability distribution. The 𝛃 includes 𝜽 as the estimated mean

value and 𝐯, the iid error, with representing the covariance matrix of the parameters. The

attributes can be specified for correlation between them, then h(. ) becomes a joint probability

density function with non-zero off-diagonal elements of reflecting those correlations. The

notation for this choice probability under a mixed logit model assuming joint distribution is:

𝑃𝑖𝑗𝑡 = exp(𝐱𝑖𝑗𝑡𝜷)

∑ exp(𝐱𝑖𝑘𝑡

𝜷)𝐽𝑘=1

h(𝜷)d(𝜷)

which can be estimated using simulated maximum likelihood methods (Train 2003). The Halton

draws, which offer a stable set of parameter estimates (Hensher and Greene 2003) were

employed at 500 draws per iteration in the simulated maximum likelihood estimator.

As in Lim, et al. (2013), we separate the utility function into observable (𝑉𝑖𝑗𝑡) and error

components, with observable specified in the following manner:

𝑉𝑖𝑗𝑡 = 𝛼′ 𝑐𝑖𝑗𝑡 + 𝜷𝑖

′𝐱𝑖𝑗𝑡.

12

In this specification, the price of the product (c𝑖𝑗𝑡) is a scalar and has a fixed parameter α to

ensure a meaningful estimated price coefficient (Lim et al. 2013, Meijer and Rouwendal 2006).

The 𝐱𝑖𝑗𝑡 are the attributes of alternative j faced by individual i in choice scenario t. The

estimated parameters in β are random and specified to have a normal distribution with correlated

attributes.

Latent Class Model

Similar to the mixed logit model, the latent class model allows for heterogeneity in consumer

tastes but relaxes the necessity to specify any distribution function across individuals (Greene

and Hensher 2003). This extension to the methodology asserts that consumer behavior depends

on both recognizable attributes and latent heterogeneity that is not observable. These latent

subgroups are implicitly separated into a set of Q classes, which may or may not be known to the

consumers themselves, but is not known within the research context. This choice probability of

individual i choosing alternative j in the choice set t and given class q is (Lim, et al, 2013):

𝑃𝑖𝑡|𝑞(𝑗 = 1) = exp(𝛼′ 𝑐𝑖𝑗𝑡 + 𝐱′

𝑖𝑡,𝑗𝜷𝒒)

∑ exp(𝛼′ 𝑐𝑖𝑗𝑡 + 𝐱′′𝑖𝑡,𝑗𝜷𝒒)𝐽𝑗=1

.

The scalar 𝑐𝑖𝑗𝑡 signifies the price and 𝐱′𝑖𝑗𝑡 the observed characteristic of alternative j and

selection t within the choice menu. In contrast to the mixed logit, the latent class model estimates

Q sets of parameters – the 𝜷𝒒 – that describe the distinctive preferences of those within each

particular class. The formulation of the model to assign class membership probability, as

described in Greene and Hensher (2003) the following form:

𝐻𝑖𝑞 = exp(𝐳′𝑖𝝀𝒒)

∑ exp(𝐳′𝑖𝝀𝒒)𝑄𝑞=1

where 𝐳𝑖 contains the characteristics of the individual i that are observable and used to

13

characterize the membership to which each belongs, including age, female, frequent or moderate

purchaser of pecans, household size, income, consumer ethnocentricity, frequent or moderate

purchaser of organic products. The 𝝀𝒒 vector contains the estimated parameter on 𝐳𝑖, with only

Q-1 sets of estimates, due to normalization of the Qth

parameter to zero for model identification

(Greene 2008). Based on 𝐻𝑖𝑞, this method calculates the portion of respondents from the sample

that belong to each class and uses a maximum likelihood procedure to produce parameter

estimates.

The selection of the number of classes (Q) cannot be obtained by a specific parametric

statistical test (Swait 1994), though there are some common information criteria used to

determine the number of classes, such as the minimum of the Bayesian Information Criterion

(BIC) or Akaike’s Information Criterion (AIC) (Pacifico and Yoo 2013, Lim et al. 2013).

Advocated in Green and Hensher (2003) to avoid inconsistency, we “tested down” from a larger

number of classes to a smaller number. Our information criteria indicated the best fit for a three-

class model.

14

Findings and Discussion

Mixed Logit Model

Parameter estimates from the mixed logit model are presented in Table 1.4 along with estimates

of average consumer willingness-to-pay (WTP) calculated as the ratio of an attribute coefficient

and the price coefficient 𝑊𝑇𝑃𝑘 = − 𝛽𝑘

𝛼𝑝𝑟𝑖𝑐𝑒 . In this study, standard errors for the WTP estimates

are produced using the Krinsky and Robb (1986) parametric bootstrap procedure and 1,000

replications (Hole 2007a). We find significant preference heterogeneity across respondents in

most categories, with the exception of those for expiration dates and antioxidant content.

Coefficient and average WTP estimates from the mixed logit model align in part with

expectations, but also reveal some findings that may be counter to the ambitions of the pecan

FMO. Consistent with theoretical expectations, the estimated coefficient for the price and the

would-not-buy options are negative and significant, indicating that consumers prefer lower

prices. As has been found in studies of other commodities, the estimated coefficient for organic

is positive and indicates that consumers on average are willing to pay a sizable premium of $2.53

per 8 oz. for organic pecans. Although production of organic pecans is still quite small in the

U.S. and faces significant challenges from pests and tree diseases in addition to lacking supply

chain and processing infrastructure, the choice experiment indicates that there is a potential

premium available in the market place if the supply hurdles can be surmounted. Similarly, we

find that consumers are willing to pay a substantial premium ($3.47) for domestic U.S. pecans

relative to imported pecans from Mexico. This echoes findings from Palma, Collart, and

Chammoun (2015) and lends support for continued documentation of country-of-origin labeling

for informing consumers of pecan geographic origin.

15

Looking at the coefficient and WTP estimates for different size and grade categories, it

appears that there may be a lack of understanding of group distinctions by consumers. For the

different size labels, consumers are willing to pay a premium for Extra Large and Jumbo, but not

for Mammoth, the largest pecan size category. Further, the estimated premium for Extra Large

($2.91) is greater than for Jumbo ($1.55). This suggests that either (a) consumers prefer pecans

that are not too small (e.g., Small or Large sizes) and not too big (e.g., Mammoth sizes) or (b)

that consumers are unfamiliar with precisely what pecan size is specified by these labels.

Combining this finding with the estimates for grades tends to suggest it is the latter – consumers

are unfamiliar with the grade and size terminology. Looking at grades, a positive WTP is

estimated for Standard ($0.80) and Fancy ($1.93), but not for the intermediate grade of Choice.

Overall, these results suggest that simply implementing industry-wide uniform size and

grade standards for pecans is incomplete without accompanying efforts to inform consumers in

order to yield the intended consequences of the pecan FMO. As DeVuyst, Lusk, and DeVuyst

(2014) found with respect to the USDA beef quality grading system, effectively communicating

quality definitions is critical to ensuring consumer satisfaction and maintaining or increasing

product demand.

A further remarkable result, although not necessarily conflicting as above, is the

estimated willingness to pay for pecans with different expiration dates. The estimated

coefficients and average willingness to pay for the two longer considered expiration dates (6 and

12 months) are both positive, indicating that consumers prefer products with a longer expiration

date than the base category of 3 months. However, the estimated WTP is greater for the 6-month

expiration ($1.82) compared to the 12-month expiration ($1.07). While it might be expected that

consumers would prefer the flexibility from more time to consume a food product it is unclear

16

from previous research from other food categories, particularly unprocessed foods like pecans,

how consumers perceive long expiration dates approaching one year. As a whole, the results

indicate that any efforts under the pecan FMO to implement different ranges of expiration dates

on pecan packaging is valued by consumers, but consideration of the longevity (from a

marketing perspective) may be warranted.

Finally, looking at the coefficients and average willingness to pay for the two proposed

health and nutrition claims, a clear superiority of the Heart Healthy over the Naturally High in

Antioxidants claim is found. On average consumers are willing to pay $2.22 more for an 8 oz.

bag of pecans with the Heart Healthy claim, but no significant result is found for the High in

Antioxidants claim. Since pecans fall under the American Heart Association’s (AHA) Heart-

Check Certification Program (www.heartcheckmark.org) and are eligible to use their widely

recognized heart healthy label, the results indicate this is a profitable marketing direction.

17

Table 1.4: Mixed logit model parameter estimates and consumer WTPa

Mixed Logit Model

Parameter Estimates Willingness to Pay Estimates

Est. Coef. S.E. Est. Coef. S.E. C.I. (95%) Price -0.59

*** (0.05) Organic 1.50

*** (0.23) $2.53*** (0.44) ($1.93, $3.06) Expires in 6 mo. 1.03

*** (0.22) $1.82*** (0.47) ($1.19, $2.38) Expires in 12 mo. 0.61

*** (0.14) $1.07*** (0.31) ($0.64, $1.45) Country of Origin 2.09

*** (0.19) $3.47*** (0.44) ($2.98, $4.09) Size – Small -0.45 (0.32) -$0.65 (0.70) ($-1.63, $0.33) Size – Large 0.21 (0.27) $0.50 (0.59) ($-0.35, $1.28) Size – Extra Large 1.62

*** (0.32) $2.91*** (0.60) ($1.94, $3.94) Size – Jumbo 0.91

*** (0.28) $1.55** (0.65) ($0.69, $2.54) Size – Mammoth 0.17 (0.28) $0.32 (0.82) ($-0.61, $1.19) Grade – Standard 0.50

** (0.25) $0.80* (0.46) ($-0.09, $1.56) Grade – Choice 0.01 (0.21) -$0.12 (0.40) ($-0.71, $0.53) Grade – Fancy 1.15

*** (0.19) $1.93*** (0.37) ($1.43, $2.50) Heart Healthy 1.29

*** (0.16) $2.22*** (0.37) ($1.73, $2.70) Naturally High in Antioxidants 0.04 (0.22) $0.11 (0.44) ($-0.67, $0.81) Would-not-buy -1.11

*** (0.29) -$2.04*** (0.80) ($-3.19, $-1.20)

SD Organic -0.80

*** (0.16) Expires in 6 mo. 0.07 (0.65) Expires in 12 mo. -0.01 (0.28) Country of Origin -1.49

*** (0.17) Size – Small 0.83

** (0.38) Size – Large 0.37 (0.36) Size – Extra Large -1.13

*** (0.29) Size – Jumbo 0.48 (0.45) Size – Mammoth 2.28

*** (0.44) Grade – Standard 0.97

*** (0.31) Grade – Choice -0.19 (0.33) Grade – Fancy -0.31 (0.33) Heart Healthy -1.09

*** (0.19) Naturally High in Antioxidants -0.09 (0.28) Would-not-buy 2.48

*** (0.23)

N 6,999 6,999 Standard errors in parentheses. Omitted category for expiration variable is in comparison to expiration three months in the

future, size categories in comparison to none on the label, grade in comparison to no grade listing, and health claims in

comparison to none on the label. aCalculations are based on WTP for an 8-oz bag of pecan halves in terms of U.S. dollars.

* p < 0.10,

** p < 0.05,

*** p < 0.01

18

Latent Class Model

The parameter estimates from the latent class model are presented in Table 1.55. As with the

mixed logit model, we present coefficient and average willingness-to-pay (Table 1.6) for the

different product attributes. For each set of estimates by class, the estimated price coefficients

for the class are significant and negative, consistent with the theory that consumers have

downward sloping demand curves. Each group possesses differing price sensitivities as well,

with the estimated coefficient much larger in magnitude for classes one and three.

Latent Class One

Latent class one, with a share of about 22 percent of consumers, is the only segment for which

the would-not-buy option is positive and significant, meaning that respondents in this group

value option 3 (no purchase) over the other choices. Despite this result, there are some attributes

that they prefer with respect to pecan purchases. This group prefers organic production,

expiration in 6 or 12 months over expiring in 3 months, jumbo size, U.S. pecans over Mexico,

fancy grade, and heart-healthy labeling over none. We find that consumers with higher

ethnocentricity scores, higher income, and frequent organic purchasers were more likely to be in

class one compared to class three, though the income effect is quite small.

Latent Class Two

Latent class two, about one-third of respondents, demonstrates its own distinct preference

pattern. This group does prefer to make a purchase over choosing not to purchase. However,

similar to class one, this group also prefers organic production, U.S. pecans, fancy grade and

heart-healthy attributes. In contrast to class one, this group does not show any preference for the

freshness attributes and values size classification preferring mammoth over small. Consumers

with higher ethnocentricity scores, frequent and moderate purchasers of pecans who regularly

19

purchase organic products are much more likely to be in class two.

Latent Class Three

The last group – comprising 45 percent of consumers – reveals yet another set of preferences.

Notably, this group is not willing to pay a premium for organic pecans. Like the other two

segments, there is a strong preference for U.S. pecans over those from Mexico, fancy grade, and

heart-healthy attributes, but their only preference in terms of freshness is for those expiring 12

months in the future. Differing from the other segments, they favor all of the larger size

classifications and uniquely showed a significant desire for the naturally high in antioxidants

attribute. Previous studies have suggested that, despite general knowledge about the nutritional

properties of tree nuts (Pawlak et al., 2009; Lombardini et al, 2008), food consumers may be

unaware of the antioxidant properties of pecans (Lillywhite, 2014).

Latent Class – Consumer Willingness to Pay

Similar to the mixed logit findings, the latent class analysis results highlight some of the

differences in the variation of WTP across individual segments with respect to attribute

preferences. Overall, the WTP estimates for U.S. over Mexico pecans, fancy-grade, and heart-

healthy claims reflect positive consumer interest in these characteristics across the groups. The

organic, expiration, size and grade classifications vary in significance across the three segments

and segment three is the only one with a significant preference for the antioxidant attribute.

Though significantly valuable to all of the segmented groups, the range across the groups

of WTP for U.S. pecans over t is from $2.21/ 8 oz. bag to $6.11. The fancy grade classification is

positive and significant across all of the groups as well, with a range of $1.23 to $4.07/ 8 oz. bag.

The last of the characteristics with positive and significant results amongst all class memberships

20

includes the valuation for the heart-healthy attribute ranging from $1.26 to $5.42/8 oz. package.

Only group three demonstrates a preference for both heart healthy and antioxidant claims.

Across the groups, results reveal that only segments one and two value the organically

produced pecan (and only two shows positive utility from purchasing pecans) and on average

class two will pay a premium of $5.66/8 oz. for organic compared to class one with $3.44/8 oz.

Consumers with higher ethnocentricity scores, frequent and moderate purchasers of pecans who

commonly purchase organic products are much more likely to be in class two.

Only consumers in class one place a high value on both longer-term expiration date

options (6 and 12 months) over the short-term of 3 months at just over $2.00/8 oz. bag for each,

yet of all the size characteristics, only the jumbo distinction appears important. In contrast, class

two places no premium on the longer-term expiration dates but values the mammoth size

category ($3.49/8 oz.) with a distinct discount on the small size ($5.61/8 oz.). Class three—the

largest group— are only concerned about the longest-term expiration category (12 months) over

the 3-month with a WTP premium of $1.31/8 oz. and most highly value all of the larger size

classifications except mammoth.

Conclusion

Long-discussed issues in the pecan industry have resulted in the recently approved FMO as a

way to resolve challenges such as marketing and uniformity of standards with the goal of

ultimately increasing consumer demand. In order to assess how consumers may respond to

forthcoming initiatives in the pecan FMO and offer strategy prioritization insights to the pecan

industry, this study assesses consumer response to the FMO marketing strategies. Mixed logit

and latent class methods reveal significant taste heterogeneity across many of the attributes, but

with overall strong preferences towards U.S. origin, organic, larger size (but not too large), a

21

designated fancy grade and heart healthy claims. However, we find evidence that consumer

response to proposed size and grade designations may not align with expectations for price

premiums in the absence of efforts to educate consumers. Despite these general results,

consumers are most willing to pay for the U.S. pecans, the geographic origin attribute that is

currently required on all pecans sold at retail outlets. We also find that consumer ethnocentric

tendencies, income, and pecan and organic product purchasing frequency play a role in defining

differences among consumer taste preferences and willingness to pay.

These findings can assist pecan industry decision-makers in developing recommendations

for the industry by anticipating the consumer response to those actions and understanding that

they are not one-size-fits-all. The assumption that marketing and promotional efforts will

generate increased demand can only be accomplished by both educating consumers on product

attributes and ensuring consumer expectations for pecan quality are fulfilled. Assessing

consumer preferences in the changing environment following issuance of an FMO requires an

ongoing effort and commitment from the industry to obtain desired results.

22

Table 1.5: Latent class model parameter estimates

Class 1 Class 2 Class 3

Est. Coef. S.E. Est. Coef. S.E. Est. Coef. S.E. Price -0.45*** (0.09) -0.23*** (0.06) -0.57*** (0.07)

Organic 1.54*** (0.43) 1.44*** (0.35) 0.24 (0.43)

Expires in 6 mo. 0.98*** (0.33) 0.34 (0.42) 0.51 (0.44)

Expires in 12 mo. 1.05*** (0.36) 0.11 (0.18) 0.74*** (0.22)

Country of Origin 1.95*** (0.34) 1.45*** (0.20) 1.27*** (0.22)

Size – Small -0.23 (0.61) -1.49*** (0.57) 0.31 (0.48)

Size – Large 0.25 (0.41) -0.31 (0.71) 0.76* (0.45)

Size – Extra Large 0.43 (1.18) 0.84 (0.63) 1.09* (0.62)

Size – Jumbo 0.90* (0.50) -0.33 (0.44) 1.47*** (0.44)

Size – Mammoth -0.11 (0.53) 0.82*** (0.30) 0.10 (0.55)

Grade – Standard 0.23 (0.50) 0.88 (0.54) -0.31 (0.38)

Grade – Choice -0.50 (0.33) 0.19 (0.45) 0.31 (0.37)

Grade – Fancy 0.96** (0.37) 0.96*** (0.26) 0.69** (0.33)

Heart Healthy 0.75*** (0.29) 1.27*** (0.21) 0.88*** (0.22)

High in Antioxidants -0.04 (0.36) -0.18 (0.41) 0.75* (0.42)

Would-not-buy 2.04*** (0.51) -1.18* (0.60) -1.64*** (0.48)

Latent Segment Parameter Estimates h(.) Constant -2.18** (1.04) -4.17*** (1.12) -

Age 0.01 (0.02) 0.01 (0.02) -

Female 0.14 (0.42) 0.45 (0.39) -

Income 0.00** (0.00) 0.00* (0.00) -

Household size -0.11 (0.13) 0.14 (0.10) -

Ethnocentricity 0.54** (0.25) 0.78*** (0.25) -

Frequent pecan purchaser 0.77 (0.56) 1.04* (0.58) -

Moderate pecan purchaser 0.35 (0.51) 1.19** (0.54) - Frequent organic

purchaser 0.95* (0.55) 1.15** (0.53)

Moderate organic

purchaser -0.16 (0.46) 0.06 (0.42)

Class Probability 0.22 0.33 0.45 LL -1576.4

Standard errors in parentheses. Omitted category for expiration variable is in comparison to expiration three months in the

future, size categories in comparison to none on the label, grade in comparison to no grade listing, and health claims in

comparison to none on the label. *

p < 0.10, **

p < 0.05, ***

p < 0.01

23

Table 1.6:. Latent class model estimates of consumer WTP

Latent Class 1 Latent Class 2 Latent Class 3

Would-not-buy option $5.10 *** -$5.30 ** -$2.28 ***

($1.64, $10.51) (-$13.87, -$0.68) (-$5.03, -$0.45)

Organic $3.44 *** $5.66 *** $0.77

($1.25, $5.62)

($3.72, $8.00)

(-$1.39, $2.23)

Expiration: 6 months $2.29 *** $1.15

$1.28

($0.77, $4.11)

(-$2.63, $4.99)

(-$0.82, $2.76)

Expiration: 12 months $2.45 *** $0.49

$1.31 ***

($0.69, $4.88)

(-$1.40, $1.76)

($0.46, $2.17)

Country of origin $4.77 *** $6.11 *** $2.21 ***

($2.12, $8.92)

($4.19, $10.65)

($1.35, $3.11)

Size: Small -$1.01

-$5.61 *** $0.55

(-$4.15, $2.73)

(-$11.45, -$0.97)

(-$1.13, $2.87)

Size: Large $0.18

-$0.39

$1.20 *

(-$1.93, $2.56)

(-$7.26, $6.41)

(-$0.40, $3.38)

Size: X Large $0.33 $4.12 $1.79 *

(-$7.14, $5.31) (-$1.78, $8.76) (-$0.40, $3.61)

Size: Jumbo $1.60 * -$0.41 $2.37 ***

(-$0.86, $4.95) (-$4.11, $4.29) ($0.78, $4.74)

Size: Mammoth -$0.20 $3.49 *** -$0.04

(-$3.07, $2.46) ($0.81, $8.51) (-$1.96, $1.78)

Grade: Standard $0.97

$3.51 * -$0.43

(-$2.12, $3.41)

(-$1.00, $7.79)

(-$2.15, $0.80)

Grade: Choice -$1.00 $1.16 $0.47

(-$2.99, $0.82) (-$2.67, $5.63) (-$0.69, $1.84)

Grade: Fancy $2.36 ** $4.07 *** $1.23 **

($0.51, $4.23) ($2.50, $6.28) ($0.12, $ 2.29)

Heart Healthy $1.75 *** $5.42 *** $1.26 ***

($0.52, $3.17)

($3.45, $9.65)

($0.24, $2.74)

High in Antioxidants $0.08

-$1.01

$1.39 *

(-$2.16, $1.90) (-$4.64, $3.64) (-$0.07, $3.17) Confidence intervals in parentheses. Omitted category for expiration variable is in comparison to expiration three months

in the future, size categories in comparison to none on the label, grade in comparison to no grade listing, and health claims

in comparison to none on the label. Calculations are based on WTP for an 8-oz bag of pecan halves in terms of U.S.

dollars.

* p < 0.10,

** p < 0.05,

*** p < 0.01

24

References

American Pecan Board. 2016. The Results Are in and the Pecan FMO Passes [Press Release].

www.pecanboard.com

American Pecan Board. 2017. "Pecan FMO." accessed April 2017. http://pecanboard.com/pecan-

fmo/

Boxall, Peter C, and Wiktor L Adamowicz. 2002. "Understanding Heterogeneous Preferences in

Random Utility Models: A Latent Class Approach." Environmental and Resource

Economics 23 (4):421-446.

Boys, KA, and SC Blank. 2016. "The Evolution of Local Foods: A Retrospective and

Prospective Consideration." The Meaning of Local Foods: A Food Marketing

Perspective. J. Stanton and M. Lang, eds. Forthcoming.

Colson, Gregory, Jay R. Corrigan, Carola Grebitus, Maria L. Loureiro, and Matthew C. Rousu.

2016. "Which Deceptive Practices, If Any, Should Be Allowed in Experimental

Economics Research? Results from Surveys of Applied Experimental Economists and

Students." American Journal of Agricultural Economics 98 (2):610-621.

Costanigro, Marco, Stephan Kroll, Dawn Thilmany, and Marisa Bunning. 2014. "Is It Love for

Local/Organic or Hate for Conventional? Asymmetric Effects of Information and Taste

on Label Preferences in an Experimental Auction." Food Quality and Preference 31:94-

105.

Deselnicu, Oana C, Marco Costanigro, Diogo M Souza-Monteiro, and Dawn Thilmany

McFadden. 2013. "A Meta-Analysis of Geographical Indication Food Valuation Studies:

What Drives the Premium for Origin-Based Labels?" Journal of Agricultural and

Resource Economics:204-219.

Devuyst, Ea, Jl Lusk, and Ma Devuyst. 2014. "The USDA Quality Grades May Mislead

Consumers." Journal of Animal Science 92 (7):3142-3148.

Federal Register, Vol 81, No. 150. 2016. Final Rule, Pecan Marketing Order. In 7 CFR Part 986,

edited by Agricultural Marketing Service United States Department of Agriculture.

Florkowski, Wojciech J, Zhikang You, and Chung L Huang. 1999. "Consumer Selection of

Retail Outlets in Buying Pecans." Journal of Food Distribution Research 30:34-43.

Gold, Michael, Mihaela M Cernusca, and Larry Godsey. 2004. "Consumer Preferences for

Chestnuts, Eastern Black Walnuts, and Pecans." HortTechnology 14 (4):583-589.

Greene, William H. 2008. Econometric Analysis. Sixth ed. New Jersey: Prentice Hall.

http://www.pecanboard.com/

http://pecanboard.com/pecan-fmo/


25

Greene, William H, and David A Hensher. 2003. "A Latent Class Model for Discrete Choice

Analysis: Contrasts with Mixed Logit." Transportation Research Part B:

Methodological 37 (8):681-698.

Hensher, David A. 2010. "Hypothetical Bias, Choice Experiments and Willingness to Pay."

Transportation Research Part B: Methodological 44 (6):735-752.

Hensher, David A, and William H Greene. 2003. "The Mixed Logit Model: The State of

Practice." Transportation 30 (2):133-176.

Hole, Arne Risa. 2007a. "A Comparison of Approaches to Estimating Confidence Intervals for

Willingness to Pay Measures." Health Economics 16 (8):827-840.

Hole, Arne Risa. 2007b. "Estimating Mixed Logit Models Using Maximum Simulated

Likelihood." Stata Journal 7 (3):388-401.

Kafle, Achyut, Stephen K Swallow, and Elizabeth C Smith. 2015. "Does Public Funding Affect

Preferred Tradeoffs and Crowd-in or Crowd-out Willingness to Pay? A Watershed

Management Case." Environmental and Resource Economics 60 (3):471-495.

Klain, Tyler J., Jayson L. Lusk, Glynn T. Tonsor, and Ted C. Schroeder. 2014. "An

Experimental Approach to Valuing Information." Agricultural Economics 45 (5):635-

648.

Krinsky, Itzhak, and A Leslie Robb. 1986. "On Approximating the Statistical Properties of

Elasticities." The Review of Economics and Statistics:715-719.

Lewis, Karen E, and Carola Grebitus. 2016. "Why Us Consumers Support Country of Origin

Labeling: Examining the Impact of Ethnocentrism and Food Safety." Journal of

International Food & Agribusiness Marketing 28 (3):254-270.

Lillywhite, Jay M, Jennifer E Simonsen, and Richard J Heerema. 2014. "Us Consumer Purchases

and Nutritional Knowledge of Pecans." HortTechnology 24 (2):222-230.

Lim, Kar H., Wuyang Hu, Leigh J. Maynard, and Ellen Goddard. 2013. "U.S. Consumers'

Preference and Willingness to Pay for Country-of-Origin-Labeled Beef Steak and Food

Safety Enhancements." Canadian Journal of Agricultural Economics 61 (1):93-118.

Lombardini, Leonardo, Tina M Waliczek, and Jayne M Zajicek. 2008. "Consumer Knowledge of

Nutritional Attributes of Pecans and Factors Affecting Purchasing Behavior."

HortTechnology 18 (3):481-488.

Lusk, Jayson L, Jason Brown, Tyler Mark, Idlir Proseku, Rachel Thompson, and Jody Welsh.

2006. "Consumer Behavior, Public Policy, and Country-of-Origin Labeling." Review of

Agricultural Economics 28 (2):284-292.

26

Lusk, Jayson L, Jutta Roosen, and John A Fox. 2003. "Demand for Beef from Cattle

Administered Growth Hormones or Fed Genetically Modified Corn: A Comparison of

Consumers in France, Germany, the United Kingdom, and the United States." American

Journal of Agricultural Economics 85 (1):16-29.

Lusk, Jayson L., and Brian C. Briggeman. 2009. "Food Values." American Journal of

Agricultural Economics 91 (1):184-196.

McFadden, Daniel. 1974. "Conditional Logit Analysis of Qualitative Choice Behavior." In

Frontiers in Econometrics, 105-142.

McFadden, Daniel. 2015. "Direct Elicitation of Indirect Preferences." Paper Society for

Economic Measurement Annual Conference (Paper 144), Paris, France, July 24, 2015.

Meijer, Erik, and Jan Rouwendal. 2006. "Measuring Welfare Effects in Models with Random

Coefficients." Journal of Applied Econometrics 21 (2):227-244.

Pacifico, Daniele, and H Yoo. 2013. "lclogit: A Stata Command for Fitting Latent-Class

Conditional Logit Models Via the Expectation-Maximization Algorithm." Stata Journal.

13 (3):625-639.

Palma, Marco A, Alba J Collart, and Christopher J Chammoun. 2015. "Information Asymmetry

in Consumer Perceptions of Quality‐Differentiated Food Products." Journal of

Consumer Affairs 49 (3):596-612.

Park, Timothy A, and Wojciech J Florkowski. 1999. "Demand and Quality Uncertainty in Pecan

Purchasing Decisions." Journal of Agricultural and Applied Economics 31 (1):29-39.

Shimp, Terence A., and Subhash Sharma. 1987. "Consumer Ethnocentrism: Construction and

Validation of the Cetscale." Journal of Marketing Research 24 (3):280-289.

Swait, Joffre. 1994. "A Structural Equation Model of Latent Segmentation and Product Choice

for Cross-Sectional Revealed Preference Choice Data." Journal of Retailing and

Consumer Services 1 (2):77-89.

Tushnet, Rebecca. 2015. "COOL Story: Country of Origin Labeling and the First Amendment."

Food & Drug LJ 70:25.

USDA, Agricultural Marketing Service. "Marketing Orders & Agreements." accessed June 26,

2017. https://www.ams.usda.gov/rules-regulations/moa

USDA, Economic Research Service. 2016a. Fruit and Tree Nuts Yearbook, 2016.

USDA, Economic Research Service. 2017. Fruit and Tree Nuts Outlook/Fts-364/April 2017.

https://www.ams.usda.gov/rules-regulations/moa

27

USDA, National Agricultural Statistics Service. 2016b. Noncitrus Fruits and Nuts 2015

Summary.

Wedel, Michel, and Wagner A Kamakura. 2000. Market Segmentation: Conceptual and

Methodological Foundations: Springer Science & Business Media.

Wells, Lenny. 2017. Pecan: America's Native Nut Tree: University of Alabama Press.

Wood, Bruce W, Jerry A Payne, and Larry J Grauke. 1990. "The Rise of the Us Pecan Industry."

HortScience 25 (6):594-723.

28

CHAPTER 2: USING EXPERIMENTAL AUCTIONS TO DIRECTLY MEASURE THE

VALUE OF INFORMATION IN PECAN PURCHASES

Introduction

Product labels serve a variety of roles critical to both producers and consumers. For producers,

labels are a mechanism for a statement of quality (Kiesel, McCluskey, and Villas-Boas 2011),

product differentiation, or cost-effective access to niche markets and are a key avenue for

obtaining a price premium in crowded and competitive markets (Deselnicu et al. 2013, Bramley,

Biénabe, and Kirsten 2009). If labels are “credible, truthful, and understandable” to consumers,

they can deliver essential details about products that they can’t verify for themselves (Kuchler et

al. 2017), lessening any potential information asymmetries that could result in market failure

(Akerlof 1970). Labels may also serve to promote or define other objectives, such as social or

environmental causes, that are valued by the consumer (Golan et al. 2001). Effective labeling

makes for better-informed consumers and promotes a closer alignment of available choices with

preferences to maximize utility (Lusk et al. 2006, Kiesel, McCluskey, and Villas-Boas 2011,

Kuchler et al. 2017), with potential economic benefits for reconciling consumer misconceptions

(McFadden and Huffman 2017).1

In the context of food products, there is an enormous literature exploring consumer

preferences and willingness to pay (WTP) for products with different attributes signaled by

front- or back-of-package labels. Frequently explored labels that have consumer, producer, and

1 Though this section is focused on producers and consumers, third-party organizations and governments also have a role in the significance of labels in the market, both affecting and reflecting producer and consumer desires. See (Golan et al. 2001) for background and (Kuchler et al. 2017) for a current review of thirty years of government intervention in labeling.

29

policy implications include consumer WTP for country of origin (Loureiro and Umberger 2003,

2005, 2007, Ehmke 2006, Lim et al. 2013, Mabiso et al. 2005, Lim et al. 2014, Gao, Schroeder,

and Yu 2010, Lewis and Grebitus 2016), genetically modified foods (Lusk et al. 2005,

Dannenberg 2009, Colson and Rousu 2013), and organic (Loureiro and Hine 2002, Yue and

Tong 2009, Olesen et al. 2010, Van Loo et al. 2011). A significantly smaller literature focuses

explicitly on the value of information (VOI) – the difficult task of measuring the benefits (Golan

et al. 2001) consumers derive from the information itself conveyed by a label or information

dissemination campaign (Klain et al. 2014). Examples of VOI studies include estimates of the

value of nutritional label information (Teisl, Bockstael, and Levy 2001), value of mandatory

labeling policy for organic and cloned milk (Brooks and Lusk 2010), VOI under different

genetically modified (GM) food labeling contexts (Hu, Veeman, and Adamowicz 2005) and the

value of third-party information dissemination for food products that might be GM (Rousu et al.

2007) or comparing natural and organic claims (McFadden and Huffman 2017). Loureiro and

Umberger (2007) measure the U.S. consumer VOI for country of origin labeling, traceback

systems, food safety-related labeling programs, and product tenderness for beef ribeye steaks.

Klain et al. (2014), analyzing fresh beef and pork products in the context of mandatory country

of origin labeling, proposes a method to directly measure consumer VOI using experimental

methods, built upon the foundation and more indirect methods of Foster and Just (1989) and

Leggett (2002). Their proposed direct method involves explicitly asking consumers how much

they value knowing country of origin information versus not knowing, even if the specific

location is unidentified.

With few exceptions, the aforementioned studies have a common feature – the VOI is

estimated primarily for a single product attribute or labeling context absent other characteristics

30

listed on the label. For example, in the Rousu et al. (2007) study, they estimate the value of GM

information by comparing experimental auction bids on a plain-labeled product (e.g., 5 lb.

Russet Potatoes) and a product with a single additional label statement (“This product is made

using genetic modification (GM).”). On average, their bidders discount GM labeled products by

14% relative to those with a plain label. Participants place an additional value of about 2% of the

purchase price for knowing verifiable information about the food products that could be GM.

Similarly, Klain et al. (2014) explores the value of country of origin information by comparing a

plain-labeled meat product with a meat product bearing the country of origin. In this case the

average direct VOI estimate of “knowing vs. not knowing” the country of origin information is

$1.37 per steak, or about $0.69 per pound of meat purchased.

In this study we discuss and present evidence that these approaches to assessing the VOI

may not deliver the appropriate valuation estimates needed for policy decisions for a simple

reason – the value of information is not independent, rather it depends upon the entire

information environment. That is, the value of information regarding a single product attribute is

conditional on the other known attributes (e.g., other label signals) for that product. For products

with voluntarily provided information by producers/manufacturers or mandatory information

required under regulatory policies, estimates of the VOI may be under- or over-estimated when

evaluated in a sterile environment where these informational elements are stripped away.

To illustrate, consider an extreme example of the value of country of origin (COO)

information. Comparing consumer values between a plain-labeled product and an otherwise

equivalent product with the addition of COO information, previous studies have shown that the

average consumer has a non-trivial positive value for this information (e.g., $1.37/per steak or

chop in Klain et al. (2014)). However, consider the same exercise, but suppose both products are

31

required to bear an additional piece of information entirely unrelated to COO –“Death from

Within: Certified to contain a lethal level of cyanide”. Presumably, if research participants

notice the label (a questionable assumption as shown in Noussair, Robin, and Ruffieux (2002)),

the value for the product with or without COO information will be zero and as a result the value

of COO information in this setting will no longer be positive, but zero. Although extreme, this

example illustrates that the value of information is not independent of the entire information

environment. Further complicating the issue, as we explore in this study, it is not clear in even

simple cases whether estimates of the VOI of a particular product attribute are over- or under-

estimated when other product information is stripped away in the research experiment

environment. To illustrate this issue, consider the same exercise to estimate the value of COO

information, but instead of cyanide warnings, suppose the two products also bear an organic

label. A priori, we have no strong intuition as to why consumers would value country of origin

information the same when considering two products with or without organic information. Nor

is it clear whether consumers value COO information less when considering organic products

(perhaps because knowing the products satisfied organic standards reduces the value of origin

information) or the value of COO information is greater because consumers do not have faith in

the compliance with organic standards for products originating from certain origins.

To explore the relationship between the value of information for a product attribute when

considered in isolation against a richer information environment, we present evidence from a

series of experimental auctions modeled after the recent methodological developments of Klain

et al. (2014). In the experiments we consider the value of country of origin information in a

plain-label setting and iterative settings where additional voluntary information is provided on

the label (organic information and expiration dates). Our commodity focus is pecans, a product

32

with a recent policy emphasis surrounding the 2016 approval of a Federal Marketing Order

(FMO) for the 15-state U.S. pecan production region.2 Initiated as industry-driven agreements,

FMOs are individually tailored to promote an industry and cooperatively address its customized

needs, ultimately becoming binding regulation after approval by producers and the Secretary of

Agriculture (USDA). The 2016 FMO was obtained by the pecan industry citing challenges that

include supply concerns, prices, market disruption, and lack of uniform quality, container, and

packaging standards (American Pecan Board 2017). The final rule encompasses several actions

to be undertaken to address these issues, including collection of industry data on production,

inventory, and supply, authorization of funding for research on health and nutrition aspects of

pecans and improved technology, as well as recommendations for uniform grade, size, quality,

and container standards (Federal Register 2016).

In general, pecans are largely an understudied commodity in terms of consumer

valuation. The primary emphasis of available research includes a focus on native versus

improved pecan varieties (Palma, Collart, and Chammoun 2015), pecan consumer demographics,

consumption frequency, purchase patterns and overall tree nut nutrition knowledge (Lillywhite,

Simonsen, and Heerema 2014, Lombardini, Waliczek, and Zajicek 2008), festival attendees

preferences among three tree nuts (Gold, Cernusca, and Godsey 2004), quality perception (Park

and Florkowski 1999) and factors affecting retail outlet selection in purchasing pecans

(Florkowski, You, and Huang 1999).

2 The FMO covers pecans grown in Alabama, Arizona, Arkansas, California, Florida, Georgia, Kansas, Louisiana, Mississippi, Missouri, New Mexico, North Carolina, Oklahoma, South Carolina, and Texas. Any financing required to carry out directives of the FMO are handled through assessments on the handlers of pecans grown in these states and locally managed, under USDA oversight, by the American Pecan Council which is assembled as a result of the order. (USDA 2016) For more detail, see the final rule, effective August 5, 2016 (Federal Register 2016).

33

Further unexplored is how consumers might respond to country of origin (COO) labeling

information related to pecans. The mandatory point-of-purchase COO labeling for beef and pork

has recently been discontinued in the U.S. following legal challenges, yet the requirement for

pecans and other “covered commodities” sold at retail remain in place as a regulatory policy with

the aim of better-informing consumers (USDA). True to the stated intent of the regulation, some

studies have found that COO labeling may help to mitigate the search costs for consumers who

prefer domestic food products and functions as a solution to asymmetric information (Lusk et al.

2006). However, this information may also be perceived as a quality or food safety marker (Lim

et al. 2013, Lewis and Grebitus 2016), a way to connect with or support one’s region or locality

(Boys and Blank 2016) reflect consumer ethnocentrism (Lusk et al. 2006, Lewis and Grebitus

2016, Klain et al. 2014), represent a freedom-of-speech issue (Tushnet 2015) or provide other

signals about the product (Lusk and Briggeman 2009). Though the consumer decision-making

process can be complex (Lusk et al. 2006, Deselnicu et al. 2013, Costanigro et al. 2014) and

most studies are context-or product-specific, there is limited evidence that consumers generally

favor labeling identifying the country of origin (Lim et al. 2013), but any such substantiation for

pecan purchases requires further exploration.

In the remainder of the paper, we detail the recruitment and data collection, research

experiment design, experimental procedures, and direct VOI approach. Following , estimates of

consumer VOI are presented employing alternative econometric specifications. Finally, we

conclude with a discussion of results and implications.

34

Methods and Procedures

Recruitment and Data Collection




Consumer Lab (University of Georgia Griffin campus) database of adults in the general public.

Participants were recruited and pre-screened for tree nut allergies3 using an online survey for the

one-hour sessions. Following the screener check, participants were contacted and chose from

offered sessions, with several being held over a three-day period at different time slots with a

target of approximately twenty participants per session. Each person was reminded once, by

either email or phone, the day before the scheduled research would take place.






would take place. In both locations, each group of interested respondents was informed that they

would take part in research about consumer preferences for tree nuts and given a participation

fee of $50 upon arrival for a research session. Each participant was given a questionnaire that

included basic demographics (gender, education, age, race, and income), pecan consumption and

buying habits, awareness of mandatory country of original laws with respect to pecans, label

reading behavior with common purchases, and questions that allow a calculation of an

3 Though there would be no consumption of pecan products on-site, research was restricted only to those without nut allergies in an abundance of caution for the well-being of participants and this method was how the IRB approved the study.

35

ethnocentrism measure as outlined in Klain et al. (2014).4 The ethnocentrism measure—a 1 to 5

scale with higher indicating more—addresses the contention that differences in consumer

valuation by country of origin are driven merely by consumer protectionism preferences.

Table 2.1: Summary Statistics (N=164) Variable Description Sample Mean Std. Deviation Gender 1 = Female 0.62 0.49 Age Participant’s age 39.5 14.0 Income Household Income ($ 1,000 s) 57.4 40.6 Education Years of schooling 14.3 2.60 Household Size Number of persons 3.30 1.50 Consumer Ethnocentrism 1= low; 5= high 2.70 0.97 Eat pecans 1 = Very often/often/sometimes 0.82 0.38 Look for COO on food labels 1 = Always or often 0.31 0.46 Aware of mandatory pecan COO label 1 = Yes 0.17 0.37 Read nutrition labels on food 1 = Always or often 0.62 0.49 Look for food expiration labels 1 = Always or often 0.85 0.36 Purchase organic foods 1=Often or very often 0.18 0.38

The core component of each session is an experimental auction to directly measure

consumers’ value of information. The methodology, which is detailed in the following section

and in figure 2.1, is based upon the approach proposed by Klain et al. (2014) with two key

modifications. First, instead of using multiple price lists to elicit ranges of consumers’ value for

information, we utilized experimental auctions to elicit a direct estimate of WTP. Second, to

assess the impact of different information sets, we include multiple product options across

bidding rounds instead of a single one-shot decision.

Experiment Steps

4 We used the same method as Klain et al. (2014), which involved shortening the CETSCALE found in Shimp and Sharma (1987) from a set of 17 Likert-scaled question down to three questions. Their test of the original measure led to choosing the three that had the highest factor loadings in the general ethnocentrism scale, leading to development of a 3-question based average. This average was calculated based on the respondent’s answer to three questions where each individuals’ responses ranged from “strongly agree” to “strongly disagree” on a five-point Likert-scale. The exact language in the three questions as quoted from Klain were: “Americans should not buy foreign products, because this hurts American business and causes unemployment,” “It is not right to purchase foreign products because it puts Americans out of jobs,” and “A real American should always buy American made products.”

36

Figure 2.1 is a flowchart outlining the experiment steps for participants. For Step 1, upon arrival,

the participants check in, read and sign a consent form, receive their $50 participation fee, along

with a packet containing all materials required to participate in the experiment. The packet is

identified only with an ID number to preserve anonymity. All responses turned in by participants

contain only the ID number, with no individually identifying information. The packet includes

the pre-auction questionnaire mentioned above and page-by-page instructions to follow during

the one-hour session. Once seated in the lab, participants complete the basic demographics

portion of the questionnaire before proceeding to the next step, led by the monitor.

For Step 2, the session monitor informs participants that they will be engaging in an

auction of common food products and asked to refrain from communicating with others during

the process since the bid submissions are private They receive an explanation of the type of

auction to be conducted, that it is likely different than other types of auctions in which they have

previously participated, there is a total of six rounds of bidding with only one that counts (is

“binding”), and auction winners make an actual purchase of the one product selected. Further

explanation is given to participants, with instructions and written examples, quiz and practice

rounds, and precisely how the winner is chosen from the submitted bids. The auction method is

the random nth-price auction (Lusk and Shogren 2007, Rousu et al. 2007, Colson, Huffman, and

Rousu 2011)

In a random nth price auction, k bidders – for example k = 15 – confidentially submit

their non-negative bid for the pecans under consideration on the bid sheet provided in their

information packet. Once submitted, the session monitor places bids in order ranked from

highest to lowest. Next, the monitor draws the “random n” from a uniform distribution of

positive integers between 2 and k, for example n=4. This drawn “nth-price” becomes the nth

37

highest bid and n-1 individuals submitting bids higher than this win the auction, with each

paying the nth ranked price for the object (Colson 2009), not their own submitted bid price. In

our illustrative example, that would mean that the three highest bidders are the auction winners

(n (4)-1=3) and all three pay the 4th highest price.5

The random nth price auction elicitation device was chosen because of its prevalent use

in food values research (McFadden and Huffman 2017) and because it combines the advantages

of both stated and revealed preference, is demand revealing (Lusk and Shogren 2007) and is

incentive compatible because each participant has an incentive to bid their own private and true

value of the product. To better understand the demand revealing nature of the method, we build

upon the intuition of Vickrey’s classic second-price auction (Vickrey 1961), where 𝑘𝑘 bidders

have independent private values 𝜐𝜐𝑘𝑘 and submit sealed bids 𝑏𝑏𝑘𝑘 with payoffs 𝜋𝜋𝑘𝑘. Let 𝛽𝛽 denote the

nth-highest bid from the random integer, ∈ {2, 3,} and nature selects 𝜐𝜐𝑘𝑘 and 𝑛𝑛.

𝜋𝜋 𝑘𝑘 = �𝜐𝜐𝑘𝑘 − 𝛽𝛽 𝑖𝑖𝑖𝑖 𝑏𝑏𝑘𝑘 > 𝛽𝛽0 𝑖𝑖𝑖𝑖 𝑏𝑏𝑘𝑘 ≤ 𝛽𝛽

This payoff structure demonstrates how the dominant strategy for each bidder is to bid their own

private value for the product, or 𝑏𝑏𝑘𝑘 = 𝑣𝑣𝑘𝑘.

As opposed to other auctions, in which there is only one auction winner, the random nth

price auction provides an opportunity for n-1 bidders to win the auction and purchase the

product. Further, as opposed to other demand-revealing auctions (e.g. Vickrey or Becker,

DeGroot, and Marschak (1964)), this method better engages the off-margin bidders with low

values for the product being auctioned (Rousu et al. 2007) because they still have positive

probability of being an auction winner. McFadden and Huffman (2017) find evidence of more

5 As mentioned in Rousu et al. (2007), if there are ties present in the ranking, it can result in the number of winners being greater than n-1. This allows the advantage of these bidders receiving product at a price lower than their bid.

38

sincere bidding from low-value individuals under this method. Another advantageous property

is use of a random endogenously determined market-clearing price (Shogren et al. 2001). The

attributes of this method provide a good fit for the study objectives.

Step 3 begins with the monitor placing fictional nth price auction bids on a whiteboard

visible to all in the front of the room, demonstrating the ranking process and how auction

winners are determined. A quiz follows with answers to ensure a thorough grasp of how the

auction proceeds. Participants then engage in two practice nth price auction rounds, first using

black bean soup followed by chicken broth in order to gain experience with familiar products in

a hypothetical practice round.

Figure 2.1: Details of experiment steps

Following the framework of Klain et al. (2014) for the direct VOI method, this step

introduces practice for the choice of bidding between products in Container A and Container B

(two separate opaque boxes containing the pecans up for bid) as in the actual experiment.

39

Specifically, participants are asked to suppose that each will receive (hypothetically, for practice

rounds) a free 15 oz. can of black bean soup (or chicken broth in practice round 2) from inside of

Container A (not visible) for participating in our research. Then on the practice bid sheets, each

bids the highest price they are willing to pay to exchange the free can of black bean soup for one

from Container B about which they have additional information. Participants are not allowed to

view the product within the container, instead shown details about what information is present on

the label of the product within each container, consistent with obtaining their VOI (See figure 2.2

for example of direct VOI experiment task for the practice round). For each product, monitors

collect the bids, and rank in order from highest to lowest bid. From there, two random selections

are made. First, a coin toss to determine which of the two rounds will be considered the binding

round, followed by a random draw to select the nth price. Following the practice rounds,

monitors answer any remaining questions about the auction procedures.

40

Figure 2.2: Example of practice round direct VOI experiment task

41

Step 4 begins the experimental auction portion of the study. The experiment entails a

choice between keeping the free bag of pecans, offered to all participants, in Container A with

less information or bidding to upgrade to a bag from Container B, in which there is additional

information. Participants are shown a bag of pecans without labels for familiarity with the

product, but not allowed to look inside the container, making bids based on signs on each

displaying what is on the label of the product within. The experiment consists of six consecutive

label treatments6 in rounds of bidding, randomized in different sessions to avoid sequencing

effects (Colson, Huffman, and Rousu 2011). Prior to bidding for each label treatment option,

monitors place the two containers on a table, displaying Container A and B side-by-side, each

clearly displaying product information but with actual pecan bags inside, not visible to

participants. Container categories of information include combinations of no information,

organic information, country of origin information (United States (U.S.) or Mexico), and

expiration date (whether or not expiration date is on the label). For example, if faced with

bidding on product from Container B that includes country of origin information, participant

knows that the product contained within includes country of origin details but does not know

which country in particular. In this manner, we elicit participant VOI for knowing country of

origin information, but not for any particular country. See Table 2.2 for the available label

treatment options in each bidding round and figure 2.3 for an example auction bidding and

instruction sheet for one round. Bid sheets are collected, with no posting of bids between bidding

rounds in order to avoid potential influence on subsequent bids (Corrigan and Rousu 2011).

6 Rounds were assigned a color rather than a number for logistical ease in randomizing each auction session. Following the auction, the colors were categorized as numbers for analysis and description.

42

Table 2.2: Container label options for auction

Treatment/Bid # Container A Label Container B Label 1 None COO 2 None Organic 3 Organic Organic, COO 4 Not Organic Not Organic, COO 5 Organic, Expiration Date Organic, COO, Expiration Date 6 Not Organic, Expiration Date Not Organic, COO, Expiration Date

Note: Country of origin (COO) options in the script for bidders are described as product of the U.S., Mexico, or a combination of the two. The description of expiration dates did not include specific time frames, just explanation to bidders that the expiration date of the product would be on the label.

For Step 5, as practiced earlier, bids are sorted and ranked from highest to lowest and a

random draw selects which round will be binding as well as the random n, then the winning bid

ID numbers are posted on the whiteboard as auction winners, along with the nth price that will be

paid by the n-1 winners.

Step 6 concludes the experiment to elicit direct VOI and all participants answer the post-

auction questionnaire. Upon completion, auction winners exchange money for the bag of pecans

from Container B – the binding round label treatment, while the non-winners receive their free

product from Container A. All options presented in label treatments were available for auction

winners, following Colson et al. (2016) and (Rousu et al. 2015) to ensure that no deceptive

practices were employed.

43

Figure 2.3:. Example of auction label treatment round bid sheet

44

Direct Value Approach

Incorporating and refining the developments of Klain et al. (2014) for directly measuring VOI or

willingness to pay for information, we elicit participant bids for trading up from an endowment

of an 8 oz. bag of pecans with a particular set of information in Container A to one with more

information on the label in Container B. The participant’s bid represents the VOI, varying across

label information treatments.

Multiple regression analysis provides a manner of accounting for potentially confounding

factors, including the participant’s unique socioeconomic characteristics or for censoring of bids.

The method theoretically holds constant the participants taste between pecans in the two

containers, where the dependent variable is simply the difference between their values for pecan

products in Container A and Container B – in this case the value of information between the two

differently labeled products. Consistent with this approach, the participant’s inverse demand

equation for pecans is (Rousu et al. 2007):

𝑃𝑃𝑖𝑖𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝐶𝐶𝐶𝐶𝐶𝐶 𝐵𝐵− 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝐶𝐶𝐶𝐶𝐶𝐶 𝐴𝐴 = 𝛽𝛽1 + 𝛽𝛽2𝑋𝑋𝑖𝑖2 + 𝜀𝜀𝑖𝑖,

where 𝑃𝑃𝑖𝑖is the price bid by participant i in a particular round of bidding, 𝛽𝛽1 is the intercept, 𝑋𝑋𝑖𝑖2 is

a vector of explanatory variables for each individual, 𝛽𝛽2 is a vector of coefficients, and 𝜀𝜀𝑖𝑖 is the

error term. In this specification, the experiment design results in a bid that is the direct VOI, in

terms of dollars and cents for selecting pecans from Container B over those in Container A.

Data Summary and Results

In this section we present summary statistics of auction bid prices for information across

the six label treatments in bidding rounds and conduct unconditional tests of (a) whether there is

a value for information in the context of pecans and (b) whether the value of information is

affected by the presence of other known information on pecan labels. Table 2.3 summarizes the

45

mean bid for each of the six different label treatments. As can be seen in the table, participants

had a positive value for information across all six treatments. For example, for label treatment 1,

where participants were offered a bag of pecans with no information regarding the country of

origin (COO) of the product vs. the opportunity to bid to upgrade to a bag that would reveal the

COO of the pecans, consumers on average had a value of $1.37 for COO information. The

lowest average value of information was measured for label treatment 3 for pecans known to be

organic, but to upgrade to know the COO information ($1.32). The highest average value of

information was found in label treatment 5 for organic pecans bearing an expiration date but

unknown geographical origin vs. an otherwise equivalent bag of pecans with a known

geographical origin ($1.61).

While on average we find a positive willingness to pay for information across all six label

treatments, a significant proportion (between 55-64%) of participants submitted a zero bid

depending upon the treatment, indicating they have zero (or possibly negative) value of COO

information in all considered information settings.7 While in itself this is an interesting result

suggesting that COO information is of no value to the majority of consumers, it poses a

challenge for statistical testing of the core question of whether the value of information is

affected by the availability of other known information.

7 Overall, of the 164 participants in this experimental auction, 49 bid zero across all label treatment options.

46

Table 2.3: Summary statistics of auction bid prices by label treatment options

Summary statistics by label treatment options * All Bids (N=164) Bids > 0 Bid Container A Container B Mean % zeros Mean N 1: None COO $1.37 54.9% $3.04 74 ($2.08) ($2.12)

2: None Organic $1.46 61.0% $3.44 64 ($2.50) ($2.97)

3: Organic Organic, COO $1.32 64.0% $3.37 59 ($2.32) ($2.77)

4: Not Organic Not Organic, COO $1.42 58.0% $3.11 69 ($2.45) ($2.95)

5: Organic, Expiration Organic, COO, Expiration $1.62 61.0% $3.83 64 ($2.69) ($3.11)

6: Not Organic, Expiration Not Organic, COO, Expiration $1.61 57.0% $3.43 71 ($2.60) ($3.00)

*Standard deviation in parentheses.

In table 2.4, the difference in bids between label treatments is presented, summarizing all

of the bids (including zeros in the mean8). For example, the fourth value ($0.18) is the

difference between bid #5 (organic and expiration information vs. organic and expiration with

COO information) and bid #1 (no information vs. COO information). While it is found, for this

example, that on average consumers bid $0.18 more for COO information when both organic and

expiration date information is available than when there was no other information available

about the pecans (i.e., suggesting an increase in the value of origin information when organic and

expiration are known), the difference is not statistically different from zero (paired t-test) at

conventional significance levels. Further, for all of these bid differences, none are statistically

different from zero.

8 Though zero bids are included in these statistics, excluded are the bids from participants who bid zero for every label option.

47

Table 2.4: Mean bid difference between pairs of label treatments Mean bid difference between pairs of label treatments (N = 115)* Bid Container A Container B Bid 2 Bid 3 Bid 4 Bid 5 Bid 6

1: None COO -$0.04 -$0.23 -$0.09 $0.18 $0.16 (0.23) (0.21) (0.16) (0.23) (0.24)

2: None Organic -$0.19 -$0.05 $0.22 $0.20 (0.29) (0.21) (0.25) (0.29)

3: Organic Organic, COO $0.14 $0.40 $0.39 (0.26) (0.29) (0.30)

4: Not Organic Not Organic, COO $0.26 $0.25 (0.21) (0.27)

5: Organic, Expiration Organic, COO, Expiration -$0.01 (0.25)

6: Not Organic, Expiration Not Organic, COO, Expiration *Standard errors in parentheses. This difference between bids for varying label treatments excludes those who bid all zeros, since taking the difference two zero bids reveals no information about how the participant values either the item or the information contained in the label treatment. Difference is calculated using the bid on horizontal top columns minus bid on vertical left column. In our example, the summary mean bid labeled Bid 5 at $0.18 is the calculation resulting from Bid 5 – Bid 1 and the result is positive.

To gain some sense whether the lack of consistent and significant differences in bid

prices between label information options is driven by the absence of a clear result or due to lower

power from a small sample size, particularly due to the high percentage of zero bids, summary

statistics and statistical tests are also presented for the sample of positive bids. As can be seen in

table 2.3, when considering only bids greater than zero, average bid-price for all treatments is

substantially higher ($3.04-$3.83 across the six label treatments), with only marginal increases in

standard deviations. Table 2.5 presents the mean difference in bids between the six label

treatments and statistical tests of differences from zero for the positive-only bids. First,

comparing bids in rounds 2-6 with bids in round 1, a couple of results emerge. Since the average

positive auction bid for bid #1 was the lowest of the six treatments, the bid difference reported in

table 2.5 for treatment 1 are all positive. We find that, relatively, consumers are significantly

willing to pay $0.57 (comparison of Bid 3 – Bid 1) more for COO information when no other

information is available (Bid 1) compared to when organic information is available (Bid 3).

48

Consumers are willing to pay only $0.44 more for knowing COO information when both organic

and expiration date information is available (comparison of Bid 5 – Bid 3). These results lend

some support for the hypothesis that as available information increases (i.e., the expiration date

is known) the value of origin information diminishes. However, for a similar comparison

between the pecans which were known to be not organic (comparison of Bid 6 – Bid 4), there is

no significant difference in bids when COO information is available, nor is there a clear or

consistent pattern across all of the considered comparisons among treatments.

Table 2.5: Mean bid difference between pairs of label treatments for positive bids Mean bid difference between pairs of label treatments for positive bids Bid Container A Container B Bid 2 Bid 3 Bid 4 Bid 5 Bid 6 1: None COO $0.36 $0.57 ** $0.12 $1.02 *** $0.24 (0.32) (0.26) (0.24) (0.33) (0.15)

2: None Organic $0.49 -$0.13 $0.82 *** $0.48 (0.45) (0.22) (0.30) (0.45)

3: Organic Organic, COO -$0.57 $0.44 ** -$0.30 ** (0.34) (0.17) (0.13)

4: Not Organic Not Organic, COO $0.73 *** $0.04 (0.27) (0.16)

5: Organic, Expiration Organic, COO, Expiration -$0.52 (0.40)

6: Not Organic, Expiration Not Organic, COO, Expiration *Standard errors in parentheses. Statistical significance *<0.10, ** <0.05, *** <0.01

Overall, while the results indicate on average consumers have positive value for country

of origin information, the median individual has zero value for COO information for pecans.

How consumers’ value COO information is affected by the presence of other known product

knowledge like organic production or expiration data is simply not clear from the available data

across all treatment options. While this might be partially explained by the large percentage of

zero bidders, even when focusing solely on individuals with positive values for information, a

sufficiently clear and robust conclusion cannot be drawn. At most the data suggests there is

49

some support, but not robust, that the value of country of origin information diminishes (or at

least changes) when other information about pecans is available.

As one further assessment of the potential drivers of the value of COO information across

different information contexts, we present a series of regressions looking at the influence of

socio-demographic, shopping preferences, and ethnocentrism on bid prices. For each treatment

we estimate a standard linear regression model (OLS) and a tobit model controlling for censoring

of bids at zero using the entire sample, and again for the sample excluding the participants that

bid only zeros across all treatments. We also estimate OLS for the difference between each bid

using bid 1 as the base for comparison. As can be seen in tables 2.6-2.10, few variables are

statistically significant with consistency across the models and treatments. The negative effect of

household size is the most regularly occurring significant influence on VOI across the models,

with awareness of mandatory pecan country of origin labeling the most significant positive one.

We find some evidence of a premium for frequent consumers of pecans, regular purchasers of

organic products and nutrition label readers with respect to bid 2 in the simple information

context of no information vs. organic information. In the more complex environment of bid 5,

VOI for knowing COO when both organic and expiration date information are available, we find

that ethnocentrism is a significant factor for paying a premium under only the tobit model

specification and reading nutrition labels when considering all of the bids. We do not find

consistent support for our measure of ethnocentrism influencing values for COO information.

We find no significant difference among individuals for demographic factors except for

household size, nor among those who typically read labels or look for country of origin or

expiration information. Overall, our regression analysis offers little actionable insight into the

drivers of consumer’s value for country of origin information.

50

Conclusion

Instructive and effective labels are important for producers and consumers in conveying

information of vital concern and addressing objectives from both for efficient market

transactions. We employ a direct value of information methodology in an experimental setting

to measure how consumers value knowing information about characteristics including organic

production, country of origin and expiration dates in a more complex labeling environment than

typically used in examining these values.

As expected, we found that on average consumers have a positive value for knowing

additional information when making a purchase between two options, but the median consumer

shows no value for country of origin information in pecan purchases. However, given the large

percentage of zero bidders, a clear understanding of these values did not emerge in either

unconditional or conditional analyses of the consumer bids. At most, the data suggests there is

some evidence that the value of country of origin information diminishes (or at least changes)

when other information about pecans is available.

A similar experimental design utilizing commodities with different consumer purchasing

patterns or in a field experiment setting (e.g. grocery store, farmers market) where consumers

participate in their normal purchase surroundings offer potential for future research, particularly

to deal with the zero-valuation issue. Despite limited results, the value of information in various

label contexts remains an area of consequence for economic research.

51

Table 2.6: OLS estimates: All bids Bid (1) Bid (2) Bid (3) Bid (4) Bid (5) Bid (6) Female 0.181 0.019 -0.189 0.076 -0.346 0.500 (0.337) (0.384) (0.422) (0.434) (0.454) (0.469) Age 0.010 -0.017 0.007 0.009 0.004 -0.001 (0.013) (0.014) (0.014) (0.012) (0.016) (0.019) High Income 0.046 -0.147 -0.008 0.067 -0.397 -0.214 (0.321) (0.356) (0.358) (0.351) (0.379) (0.373) Education 0.006 -0.080 0.096 -0.071 -0.074 -0.011 (0.067) (0.082) (0.085) (0.101) (0.105) (0.104) Household Size -0.234** -0.286** -0.106 -0.239* -0.287* -0.205 (0.115) (0.128) (0.115) (0.141) (0.159) (0.138) Ethnocentrism 0.252 0.066 0.177 0.076 0.264 -0.002 (0.206) (0.208) (0.222) (0.209) (0.243) (0.220) Eat Pecans -0.099 0.993*** 0.190 0.126 0.255 0.271 (0.395) (0.355) (0.413) (0.433) (0.453) (0.469) Look for COO on food labels -0.232 -0.423 0.078 0.116 -0.061 -0.464 (0.405) (0.495) (0.491) (0.510) (0.592) (0.548) Aware of mandatory pecan COO label 0.590 1.029* 0.969 0.772 1.242 1.657* (0.484) (0.607) (0.637) (0.718) (0.785) (0.930) Read nutrition labels on food 0.335 0.725 0.172 0.487 0.830 0.916* (0.363) (0.479) (0.336) (0.499) (0.503) (0.471) Look for food expiration labels 0.074 0.427 0.065 -0.198 0.602 0.184 (0.426) (0.402) (0.417) (0.475) (0.447) (0.533) Purchase organic foods 0.320 1.032* 0.760 0.247 0.201 -0.318 (0.470) (0.560) (0.578) (0.525) (0.680) (0.646) Constant 0.560 2.067 -1.122 1.955 1.521 1.119 (1.361) (1.399) (1.497) (1.822) (1.792) (1.847) N 159 159 159 159 159 159 adj. R-sq 0.014 0.094 0.026 0.005 0.042 0.036 F (12,146) 1.537 1.947** 0.751 1.117 1.541 1.203

Standard errors in parentheses * p<0.10, ** p<0.05, *** p<0.01

52

Table 2.7: Tobit estimates: All bids Bid (1) Bid (2) Bid (3) Bid (4) Bid (5) Bid (6) Female 0.320 -0.743 -0.994 0.221 -1.336 0.872 (0.714) (0.888) (0.989) (0.837) (1.069) (0.930) Age 0.026 -0.028 0.049 0.023 0.040 0.014 (0.028) (0.035) (0.039) (0.033) (0.043) (0.037) High Income -0.070 -1.161 0.147 0.004 -1.143 -0.436 (0.696) (0.863) (0.953) (0.811) (1.074) (0.906) Education 0.100 -0.111 0.263 0.048 -0.021 0.052 (0.139) (0.173) (0.193) (0.165) (0.206) (0.183) Household Size -0.490* -0.554* -0.143 -0.487 -0.574 -0.302 (0.249) (0.307) (0.347) (0.299) (0.392) (0.326) Ethnocentrism 0.641* 0.421 0.556 0.381 1.380** 0.306 (0.373) (0.461) (0.508) (0.440) (0.575) (0.490) Eat Pecans -0.357 2.564** 1.349 0.357 -0.083 0.633 (0.919) (1.272) (1.385) (1.097) (1.447) (1.235) Look for COO on food labels -0.622 -0.687 0.255 0.667 -0.061 -0.388 (0.813) (0.970) (1.087) (0.941) (1.187) (1.043) Aware of pecan COO label 0.778 1.663 1.843 0.881 2.926** 2.488** (0.866) (1.032) (1.155) (1.008) (1.232) (1.097) Read nutrition labels on food 0.504 1.830* 0.358 0.436 2.268* 1.732* (0.789) (1.003) (1.116) (0.929) (1.255) (1.037) Look for food expiration labels 0.405 0.935 0.334 -0.896 1.937 -0.032 (0.969) (1.229) (1.348) (1.091) (1.578) (1.242) Purchase organic foods 0.477 2.619** 1.924 0.585 0.606 -0.653 (0.896) (1.027) (1.182) (1.028) (1.300) (1.159) Constant -3.486 -1.525 -10.373** -2.364 -7.142 -4.030 (2.845) (3.676) (4.184) (3.344) (4.511) (3.786) Sigma Constant 3.500*** 4.099*** 4.521*** 4.031*** 4.934*** 4.525*** (0.328) (0.413) (0.479) (0.390) (0.503) (0.427) N 159 159 159 159 159 159 LR chi2 (12) 16.495 31.788*** 20.369 14.311 29.131** 14.634 p 0.170 0.001 0.060 0.281 0.004 0.262


53

Table 2.8: OLS estimates: All bids except for all-zero across treatments Bid (1) Bid (2) Bid (3) Bid (4) Bid (5) Bid (6) Female 0.364 0.025 -0.139 0.198 -0.302 0.759 (0.450) (0.485) (0.578) (0.596) (0.609) (0.598) Age 0.003 -0.032* -0.004 0.000 -0.006 -0.011 (0.016) (0.018) (0.019) (0.016) (0.020) (0.024) High Income 0.092 -0.083 0.051 0.204 -0.407 -0.197 (0.425) (0.492) (0.466) (0.454) (0.498) (0.501) Education -0.070 -0.188* 0.061 -0.182 -0.194 -0.105 (0.086) (0.112) (0.111) (0.147) (0.142) (0.134) Household Size -0.379** -0.456** -0.174 -0.420* -0.501** -0.349* (0.155) (0.185) (0.164) (0.230) (0.238) (0.186) Ethnocentrism 0.328 0.126 0.275 0.091 0.394 0.044 (0.260) (0.275) (0.317) (0.275) (0.323) (0.283) Eat Pecans -0.199 1.257** 0.151 0.193 0.172 0.305 (0.583) (0.496) (0.639) (0.627) (0.627) (0.655) Look for COO on food labels -0.246 -0.417 0.180 0.215 -0.042 -0.504 (0.470) (0.580) (0.577) (0.607) (0.684) (0.640) Aware of mandatory pecan COO label 0.739 1.362** 1.220 1.019 1.668* 1.977* (0.565) (0.685) (0.819) (0.852) (0.907) (1.072) Read nutrition labels on food 0.191 0.505 -0.045 0.336 0.759 0.894 (0.495) (0.631) (0.508) (0.596) (0.618) (0.632) Look for food expiration labels 0.024 0.266 -0.018 -0.445 0.732 -0.069 (0.611) (0.556) (0.630) (0.663) (0.632) (0.730) Purchase organic foods 0.192 1.129* 0.821 0.107 0.130 -0.621 (0.553) (0.635) (0.735) (0.612) (0.804) (0.776) Constant 2.799 5.055** 0.263 5.004* 4.427 3.836 (1.933) (2.165) (2.219) (2.936) (2.865) (2.646) N 112 112 112 112 112 112 adj. R-sq -0.015 0.106 -0.023 -0.010 0.037 0.021 F (12, 99) 1.183 2.098** 0.455 0.853 1.224 1.042


54

Table 2.9: Tobit estimates: All bids except for all-zero across treatments Bid (1) Bid (2) Bid (3) Bid (4) Bid (5) Bid (6) Female 0.491 -0.593 -0.672 0.378 -0.836 1.169 (0.659) (0.823) (0.950) (0.794) (0.998) (0.881) Age 0.007 -0.054 0.016 -0.000 0.005 -0.014 (0.026) (0.033) (0.039) (0.032) (0.041) (0.035) High Income 0.076 -0.838 0.311 0.257 -0.940 -0.220 (0.659) (0.819) (0.954) (0.791) (1.035) (0.880) Education -0.044 -0.275* 0.123 -0.139 -0.215 -0.121 (0.129) (0.163) (0.191) (0.157) (0.196) (0.175) Household Size -0.596** -0.713** -0.267 -0.637** -0.802** -0.477 (0.237) (0.300) (0.352) (0.292) (0.382) (0.320) Ethnocentrism 0.544 0.334 0.461 0.271 1.305** 0.203 (0.357) (0.448) (0.516) (0.435) (0.567) (0.480) Eat Pecans -0.332 2.588** 1.215 0.503 -0.310 0.739 (0.897) (1.261) (1.402) (1.095) (1.452) (1.238) Look for COO on food labels -0.514 -0.551 0.437 0.753 0.054 -0.342 (0.722) (0.880) (1.028) (0.866) (1.086) (0.955) Aware of mandatory pecan COO label 0.739 1.765* 1.722 0.939 2.929** 2.396** (0.784) (0.951) (1.116) (0.948) (1.160) (1.030) Read nutrition labels on food -0.005 0.842 -0.438 -0.200 1.546 1.035 (0.760) (0.972) (1.133) (0.919) (1.239) (1.025) Look for food expiration labels 0.143 0.237 -0.016 -1.245 1.743 -0.590 (0.934) (1.194) (1.367) (1.098) (1.530) (1.243) Purchase organic foods 0.112 2.300** 1.507 0.171 0.208 -1.188 (0.806) (0.942) (1.128) (0.956) (1.200) (1.073) Constant 1.765 5.247 -4.077 4.023 0.310 2.837 (2.772) (3.655) (4.232) (3.345) (4.442) (3.792) Sigma Constant 2.904*** 3.460*** 4.027*** 3.460*** 4.236*** 3.859*** (0.262) (0.337) (0.416) (0.324) (0.418) (0.351) N 112 112 112 112 112 112 LR chi2 (12) 11.521 28.670 11.100 10.374 22.363 11.347 p 0.485 0.004*** 0.520 0.583 0.034** 0.499


55

Table 2.10: OLS estimates: Bid differences* (1)

Bid 2 (2)

Bid 3 (3)

Bid 4 (4)

Bid 5 (5)

Bid 6 Female -0.340 -0.166 -0.503 -0.667 0.395 (0.490) (0.413) (0.456) (0.564) (0.539) Age -0.035* -0.002 -0.006 -0.009 -0.013 (0.019) (0.012) (0.015) (0.017) (0.021) High Income -0.175 0.112 -0.041 -0.499 -0.289 (0.499) (0.258) (0.419) (0.473) (0.363) Education -0.117 -0.112 0.132 -0.124 -0.035 (0.093) (0.099) (0.110) (0.121) (0.122) Household Size -0.078 -0.041 0.204 -0.122 0.029 (0.155) (0.154) (0.176) (0.203) (0.131) Ethnocentrism -0.202 -0.237 -0.053 0.066 -0.284 (0.246) (0.224) (0.247) (0.275) (0.207) Eat Pecans 1.456* 0.392 0.350 0.371 0.504 (0.749) (0.376) (0.455) (0.688) (0.466) Look for COO on food labels -0.171 0.462 0.426 0.204 -0.258 (0.618) (0.400) (0.472) (0.532) (0.475) Aware of mandatory pecan COO label 0.622 0.280 0.480 0.929 1.237 (0.573) (0.581) (0.634) (0.887) (1.056) Read nutrition labels on food 0.314 0.145 -0.237 0.568 0.703 (0.680) (0.421) (0.484) (0.565) (0.469) Look for food expiration labels 0.242 -0.468 -0.041 0.708 -0.093 (0.708) (0.361) (0.685) (0.771) (0.464) Purchase organic foods 0.937** -0.085 0.628 -0.063 -0.814 (0.463) (0.343) (0.729) (0.692) (0.656) Constant 2.256 2.205 -2.536 1.628 1.037 (2.289) (1.849) (2.151) (2.527) (2.090) N 112 112 112 112 112 Adjusted R2 0.067 -0.045 -0.025 -0.027 0.004 F (12, 99) 2.366*** 0.726 0.707 0.611 0.992

*Each dependent variable is measured with respect to differences from Bid 1. Standard errors in parentheses * p<0.10, ** p<0.05, *** p<0.01

56

References

Akerlof, George A. 1970. "The Market for Lemons: Quality Uncertainty and the Market Mechanism." The Quarterly Journal of Economics:488-500.

American Pecan Board. 2017. "Pecan FMO." accessed April 2017. http://pecanboard.com/pecan-fmo/

Becker, Gordon M, Morris H Degroot, and Jacob Marschak. 1964. "Measuring Utility by a Single‐Response Sequential Method." Systems Research and Behavioral Science 9 (3):226-232.

Boys, KA, and SC Blank. 2016. "The Evolution of Local Foods: A Retrospective and Prospective Consideration." The Meaning of Local Foods: A Food Marketing Perspective. J. Stanton and M. Lang, eds. Forthcoming.

Bramley, Cerkia, Estelle Biénabe, and Johann Kirsten. 2009. "The Economics of Geographical Indications: Towards a Conceptual Framework for Geographical Indication Research in Developing Countries." The Economics of Intellectual Property:109.

Brooks, Kathleen, and Jayson L Lusk. 2010. "Stated and Revealed Preferences for Organic and Cloned Milk: Combining Choice Experiment and Scanner Data." American Journal of Agricultural Economics 92 (4):1229-1241.

Colson, Gregory. 2009. "Improving Nutrient Content through Genetic Modification: Evidence from Experimental Auctions on Consumer Acceptance and Willingness to Pay for Intragenic Foods." Iowa State University (10801).

Colson, Gregory, Jay R. Corrigan, Carola Grebitus, Maria L. Loureiro, and Matthew C. Rousu. 2016. "Which Deceptive Practices, If Any, Should Be Allowed in Experimental Economics Research? Results from Surveys of Applied Experimental Economists and Students." American Journal of Agricultural Economics 98 (2):610-621.

Colson, Gregory J, Wallace E Huffman, and Matthew C Rousu. 2011. "Improving the Nutrient Content of Food through Genetic Modification: Evidence from Experimental Auctions on Consumer Acceptance." Journal of Agricultural and Resource Economics:343-364.

Colson, Gregory, and Matthew C Rousu. 2013. "What Do Consumer Surveys and Experiments Reveal and Conceal About Consumer Preferences for Genetically Modified Foods?" GM Crops & Food 4 (3):158-165.



57

Corrigan, Jay R., and Matthew C. Rousu. 2011. "Are Experimental Auctions Demand Revealing When Values Are Affiliated?" American Journal of Agricultural Economics 93 (2):514-520.

Costanigro, Marco, Stephan Kroll, Dawn Thilmany, and Marisa Bunning. 2014. "Is It Love for Local/Organic or Hate for Conventional? Asymmetric Effects of Information and Taste on Label Preferences in an Experimental Auction." Food Quality and Preference 31:94-105.

Dannenberg, Astrid. 2009. "The Dispersion and Development of Consumer Preferences for Genetically Modified Food—a Meta-Analysis." Ecological Economics 68 (8):2182-2192.

Deselnicu, Oana C, Marco Costanigro, Diogo M Souza-Monteiro, and Dawn Thilmany McFadden. 2013. "A Meta-Analysis of Geographical Indication Food Valuation Studies: What Drives the Premium for Origin-Based Labels?" Journal of Agricultural and Resource Economics:204-219.

Ehmke, Mariah Tanner. 2006. "International Differences in Consumer Preferences for Food Country-of-Origin: A Meta-Analysis." American Agricultural Economics Association Meeting, Long Beach CA.

Federal Register, Vol 81, No. 150. 2016. Final Rule, Pecan Marketing Order. In 7 CFR Part 986, edited by Agricultural Marketing Service,United States Department of Agriculture.

Florkowski, Wojciech J, Zhikang You, and Chung L Huang. 1999. "Consumer Selection of Retail Outlets in Buying Pecans." Journal of Food Distribution Research 30:34-43.

Foster, William, and Richard E Just. 1989. "Measuring Welfare Effects of Product Contamination with Consumer Uncertainty." Journal of Environmental Economics and Management 17 (3):266-283.

Gao, Zhifeng, Ted C Schroeder, and Xiaohua Yu. 2010. "Consumer Willingness to Pay for Cue Attribute: The Value Beyond Its Own." Journal of International Food & Agribusiness Marketing 22 (1-2):108-124.

Golan, Elise, Fred Kuchler, Lorraine Mitchell, Cathy Greene, and Amber Jessup. 2001. "Economics of Food Labeling." Journal of Consumer Policy 24 (2):117-184.

58

Gold, Michael, Mihaela M Cernusca, and Larry Godsey. 2004. "Consumer Preferences for Chestnuts, Eastern Black Walnuts, and Pecans." HortTechnology 14 (4):583-589.

Hu, Wuyang, Michele M Veeman, and Wiktor L Adamowicz. 2005. "Labelling Genetically Modified Food: Heterogeneous Consumer Preferences and the Value of Information." Canadian Journal of Agricultural Economics/Revue canadienne d'agroeconomie 53 (1):83-102.

Kiesel, Kristin, Jill J Mccluskey, and Sofia B Villas-Boas. 2011. "Nutritional Labeling and Consumer Choices." Annu. Rev. Resour. Econ. 3 (1):141-158.

Klain, Tyler J., Jayson L. Lusk, Glynn T. Tonsor, and Ted C. Schroeder. 2014. "An Experimental Approach to Valuing Information." Agricultural Economics 45 (5):635-648.

Kuchler, Fred, Catherine Greene, Maria Bowman, Kandice K. Marshall, John Bovay, and Lori Lynch. 2017. Beyond Nutrition and Organic Labels−30 Years of Experience with Intervening in Food Labels, ERR-239,. edited by U.S. Department of Agriculture Economic Research Service.

Leggett, Christopher G. 2002. "Environmental Valuation with Imperfect Information the Case of the Random Utility Model." Environmental and Resource Economics 23 (3):343-355.

Lewis, Karen E, and Carola Grebitus. 2016. "Why Us Consumers Support Country of Origin Labeling: Examining the Impact of Ethnocentrism and Food Safety." Journal of International Food & Agribusiness Marketing 28 (3):254-270.

Lillywhite, Jay M, Jennifer E Simonsen, and Richard J Heerema. 2014. "Us Consumer Purchases and Nutritional Knowledge of Pecans." HortTechnology 24 (2):222-230.

Lim, Kar H, Wuyang Hu, Leigh J Maynard, and Ellen Goddard. 2014. "A Taste for Safer Beef? How Much Does Consumers’ Perceived Risk Influence Willingness to Pay for Country‐of‐Origin Labeled Beef." Agribusiness 30 (1):17-30.

Lim, Kar H., Wuyang Hu, Leigh J. Maynard, and Ellen Goddard. 2013. "U.S. Consumers' Preference and Willingness to Pay for Country-of-Origin-Labeled Beef Steak and Food Safety Enhancements." Canadian Journal of Agricultural Economics 61 (1):93-118.

59

Lombardini, Leonardo, Tina M Waliczek, and Jayne M Zajicek. 2008. "Consumer Knowledge of Nutritional Attributes of Pecans and Factors Affecting Purchasing Behavior." HortTechnology 18 (3):481-488.

Loureiro, Maria L, and Susan Hine. 2002. "Discovering Niche Markets: A Comparison of Consumer Willingness to Pay for Local (Colorado Grown), Organic, and Gmo-Free Products." Journal of Agricultural and Applied Economics 34 (3):477-487.

Loureiro, Maria L., and Wendy J. Umberger. 2003. "Estimating Consumer Willingness to Pay for Country-of-Origin Labeling." Journal of Agricultural and Resource Economics 28 (2):287-301.

Loureiro, Maria L., and Wendy J. Umberger. 2005. "Assessing Consumer Preferences for Country-of-Origin Labeling." Journal of Agricultural and Applied Economics 37 (1):49-63.

Loureiro, Maria L., and Wendy J. Umberger. 2007. "A Choice Experiment Model for Beef: What Us Consumer Responses Tell Us About Relative Preferences for Food Safety, Country-of-Origin Labeling and Traceability." Food Policy 32 (4):496-514.

Lusk, Jayson L, Jason Brown, Tyler Mark, Idlir Proseku, Rachel Thompson, and Jody Welsh. 2006. "Consumer Behavior, Public Policy, and Country-of-Origin Labeling." Review of Agricultural Economics 28 (2):284-292.

Lusk, Jayson L, Mustafa Jamal, Lauren Kurlander, Maud Roucan, and Lesley Taulman. 2005. "A Meta-Analysis of Genetically Modified Food Valuation Studies." Journal of Agricultural and Resource Economics:28-44.

Lusk, Jayson L, and Jason F Shogren. 2007. Experimental Auctions: Methods and Applications in Economic and Marketing Research: Cambridge University Press.

Lusk, Jayson L., and Brian C. Briggeman. 2009. "Food Values." American Journal of Agricultural Economics 91 (1):184-196.

Mabiso, Athur, James Sterns, Lisa House, and Allen Wysocki. 2005. "Estimating Consumers’ Willingness-to-Pay for Country-of-Origin Labels in Fresh Apples and Tomatoes: A Double-Hurdle Probit Analysis of American Data Using Factor Scores." American Agricultural Economics Association Annual Meeting, Providence, Rhode Island.

60

McFadden, Jonathan R, and Wallace E Huffman. 2017. "Willingness-to-Pay for Natural, Organic, and Conventional Foods: The Effects of Information and Meaningful Labels." Food Policy 68:214-232.

Noussair, Charles, Stephane Robin, and Bernard Ruffieux. 2002. "Do Consumers Not Care About Biotech Foods or Do They Just Not Read the Labels?" Economics Letters 75 (1):47-53.

Olesen, Ingrid, Frode Alfnes, Mia Bensze Røra, and Kari Kolstad. 2010. "Eliciting Consumers' Willingness to Pay for Organic and Welfare-Labelled Salmon in a Non-Hypothetical Choice Experiment." Livestock Science 127 (2):218-226.

Palma, Marco A, Alba J Collart, and Christopher J Chammoun. 2015. "Information Asymmetry in Consumer Perceptions of Quality‐Differentiated Food Products." Journal of Consumer Affairs 49 (3):596-612.

Park, Timothy A, and Wojciech J Florkowski. 1999. "Demand and Quality Uncertainty in Pecan Purchasing Decisions." Journal of Agricultural and Applied Economics 31 (1):29-39.

Rousu, Matthew C, Gregory Colson, Jay R Corrigan, Carola Grebitus, and Maria L Loureiro. 2015. "Deception in Experiments: Towards Guidelines on Use in Applied Economics Research." Applied Economic Perspectives and Policy 37 (3):524-536.

Rousu, Matthew, Wallace E. Huffman, Jason F. Shogren, and Abebayehu Tegene. 2007. "Effects and Value of Verifiable Information in a Controversial Market: Evidence from Lab Auctions of Genetically Modified Food." Economic Inquiry 45 (3):409-432.

Shimp, Terence A., and Subhash Sharma. 1987. "Consumer Ethnocentrism: Construction and Validation of the Cetscale." Journal of Marketing Research 24 (3):280-289.

Shogren, Jason F, Michael Margolis, Cannon Koo, and John A List. 2001. "A Random nth-Price Auction." Journal of Economic Behavior & Organization 46 (4):409-421.

Teisl, Mario F, Nancy E Bockstael, and Alan Levy. 2001. "Measuring the Welfare Effects of Nutrition Information." American Journal of Agricultural Economics 83 (1):133-149.

Tushnet, Rebecca. 2015. "COOL Story: Country of Origin Labeling and the First Amendment." Food & Drug LJ 70:25.

61

USDA, Agricultural Marketing Service. n.d. "Marketing Orders & Agreements." accessed June 26, 2017. https://www.ams.usda.gov/rules-regulations/moa

Van Loo, Ellen J, Vincenzina Caputo, Rodolfo M Nayga, Jean-Francois Meullenet, and Steven C Ricke. 2011. "Consumers’ Willingness to Pay for Organic Chicken Breast: Evidence from Choice Experiment." Food Quality and Preference 22 (7):603-613.

Vickrey, William. 1961. "Counterspeculation, Auctions, and Competitive Sealed Tenders." The Journal of Finance 16 (1):8-37.

Yue, Chengyan, and Cindy Tong. 2009. "Organic or Local? Investigating Consumer Preference for Fresh Produce Using a Choice Experiment with Real Economic Incentives." HortScience 44 (2):366-371.

https://www.ams.usda.gov/rules-regulations/moa

62

CHAPTER 3: CONSUMER PREFERENCES FOR SINGLE VS. MIXED COUNTRY OF

ORIGIN: EVIDENCE FROM EXPERIMENTAL AUCTIONS ON THE INFLUENCE OF

RISK AND AMBIGUITY ATTITUDES

Introduction

Under the classical microeconomic theory of consumers it is assumed that consumers have well-

defined preferences and ceteris paribus, certainty is preferred over uncertainty. In the context of

food, supplying products that are from a single source or processed in a unique facility free from

any commingling or cross-contamination can be costly for the food industry due to the

seasonality of many ingredients and the inherent costs in ensuring the traceability and

segregation of food ingredients. Providing certain ingredients or attributes poses a dilemma for

both niche and multi-national players in the food industry.

As a result, many products either (a) explicitly state that the product is from a specific

origin, (b) may come from several origins, (c) processed in a facility with cross-contamination,

or (d) processed in a facility with possible cross-contamination (e.g., wheat, peanuts, etc.). In

this study we explore two important dimensions of consumer preferences for known vs.

uncertain product origins in the context of single and mixed origin product sources.

For producers and food manufacturers of specialty food products or with process

differentiation, securing premium pricing for particular consumer segments requires consumer

confidence in product attributes or ingredients. Moreover, for efficient economic activity, these

properties must be clearly relayed to consumers with labels that are truthful and understandable

(Kuchler et al. 2017). Any handling of food products diverging from the commodity standard to

63

meet consumer preferences reduces savings from economies of scale in bulk handling (Hayes

and Lence 2015). If multiple country sources are required in fulfilling demand, even if only

occasionally as with seasonal products, similar challenges exist for meeting consumer desires

with the certainty they prefer. However, multi-national sourcing offers the additional complexity

of meeting country of origin regulatory requirements, which are themselves controversial, with

those for mixed origin products remaining contentious (Lim et al. 2014) even after recent

rulings.1

For products covered under U.S. Federal country of origin regulations, retailers are

required to identify the country of origin, with some flexibility for format and placement as long

as it is clear to the customer. Regulations most recently updated in the 2016 Consolidated

Appropriations Act require customer notification on fresh and frozen fruits and vegetables, some

nuts (including pecans), fish and shellfish, ginseng, and certain meats and are referred to as

“covered commodities” since they are covered under the ruling. Statutorily, the statements may

not use phrases for mixed origin products that include terms such as “or,” “may contain,” or

“and/or” that suggest only the possible country of origin. The consumer must receive

unambiguous origin information in order to maintain the intent of the requirements (USDA n.d.)

and there are always tradeoffs in how information is presented to consumers, whether they may

be confusing (Kuchler et al. 2017) or misleading (DeVuyst, Lusk, and DeVuyst 2014), or

whether they even read them at all (Noussair, Robin, and Ruffieux 2002).

The specificity of the regulation increases the challenges to producers in meeting

consumer demand and fulfilling preferences simultaneously. Evidence from USDA compliance

surveillance reveals subpar adherence to country of origin requirements at the retail level, with

1 See (Greene 2016) for information about the recent World Trade Organization (WTO) trade dispute on meat labeling, which also includes a history and timeline for country of origin labeling regulations.

64

only 29% either fully or adequately compliant, with the most frequent non-compliance found for

absence of country of origin (USDA 2017). This suggests the presence of challenges making

their way through the supply chain from producer to retailer, further complicating the aim of

addressing the consumer’s right to know.

Explorations of consumer willingness to pay for country of origin (Loureiro and

Umberger 2003, 2005, 2007, Ehmke 2006, Lim et al. 2013, Mabiso et al. 2005, Lim et al. 2014,

Gao, Schroeder, and Yu 2010, Lewis and Grebitus 2016) are frequent, but there is scant evidence

on how policies about relaying multiple origin product information on the label are perceived by

consumers and their willingness to pay for those products might be affected. Tonsor, Schroeder,

and Lusk (2013) expanded the breadth of studies that typically look only at single origin labels

compared to a not labeled option, with an analysis of consumer preferences for origin

information on meat products that included multi-country, mandatory, and voluntary label

treatment options. Their findings revealed that, in comparison to unlabeled products, consumers

significantly valued the origin information but the multi-country labels were least preferred in

terms of willingness to pay. Conversely, they found that consumers were indifferent to direct

comparisons between Product of North America and Product of United States labels, where there

was not an unlabeled choice. Tonsor, Schroeder, and Lusk (2013) were first to address

alternative versions of provenance labels for meat products, noting important implications for

industry costs and compliance as well as consumer valuation, but also seeking to address broader

issues of rationales behind mandatory country of origin regulations and effects on both covered

and exempt products. In the context of willingness to pay for the value of information, Klain et

al. (2014) offers at least the possibility of multi origin meat products in a series of grocery store

field experiments. Their subjects placed bids on how much they would pay to know the country

65

of origin or a non-specific proportion of mixed origins, not for the specific origin. Results

suggest that the least preferred options were any meat products that include or may include those

from Mexico.

This study examines how the presence of multiple countries on the label, in accordance

with either existing mandatory rules or voluntary label placement might affect consumer values

in purchasing these products. In an extension of the available literature, we further explore how

consumer willingness to pay for single or mixed origin products might be influenced by their risk

and ambiguity attitudes. A better understanding of consumer preferences and willingness to pay

for single or mixed origin/attribute products and the relevance of certainty for those consumers is

important to industry and policy-makers, regardless of whether the labeling information is

mandatory and may also shed light on preferences for other multi-attribute products.

Our commodity focus is pecans, which are largely an understudied commodity in terms

of consumer valuation. The primary emphasis of available research includes a focus on native

versus improved pecan varieties (Palma, Collart, and Chammoun 2015), pecan consumer

demographics, consumption frequency, purchase patterns and overall tree nut nutrition

knowledge (Lillywhite, Simonsen, and Heerema 2014, Lombardini, Waliczek, and Zajicek

2008), festival attendees preferences among three tree nuts (Gold, Cernusca, and Godsey 2004),

quality perception (Park and Florkowski 1999) and factors affecting retail outlet selection in

purchasing pecans (Florkowski, You, and Huang 1999).

In the remainder of the paper, we detail the recruitment and data collection, research

experiment design and experimental procedures. Following are estimates of consumer

willingness to pay for label treatment options in the presence of uncertainty are presented

66

employing alternative econometric specifications. The final section summarizes findings and

implications of experiment results.

Methods and Procedures

Recruitment and Data Collection




Consumer Lab (University of Georgia Griffin campus) database of adults in the general public.

Participants were recruited and pre-screened for tree nut allergies2 using an online survey for the

one-hour sessions. Following the screener check, participants were contacted and chose from

offered sessions, with several being held over a three-day period at different time slots with a

target of approximately twenty participants per session. Each person was reminded once, by

either email or phone, the day before the scheduled research would take place.






would take place. In both locations, each group of interested respondents was informed that they

would take part in research about consumer preferences for tree nuts and given a participation

fee of $50 upon arrival for a research session. Each participant was given a questionnaire that

included basic demographics (gender, education, age, race, and income – see table 3.1 for

2 Though there would be no consumption of pecan products on-site, research was restricted only to those without nut allergies in an abundance of caution for the well-being of participants and this method was how the IRB approved the study.

67

summary statistics), pecan consumption and buying habits, awareness of mandatory country of

original laws with respect to pecans, label reading behavior with common purchases, and

questions that allow a calculation of an ethnocentrism measure as outlined in Klain et al. (2014).3

The ethnocentrism measure—a 1 to 5 scale with higher indicating more—addresses the

contention that differences in consumer valuation by country of origin are driven merely by

consumer protectionism preferences.

Table 3.1: Summary statistics (N=129)

Variable Description Sample Mean Std. Deviation Risk Aversion (CRRA coefficient) Midpoint of CRRA range -0.25 1.07 Ambiguity Aversion (Alpha) Midpoint of Alpha coeff.range 0.53 0.32 Bid certainty 1-10 scale - that bid is true WTP 5.14 2.34 Gender 1 = Female 0.62 0.49 Age Participant’s age 40.3 15.9 Income Household Income ($ 1,000s) 59.7 44.3 Education Years of schooling 14.3 2.70 Household Size Number of persons 2.88 1.45 Consumer Ethnocentrism 1= low; 5= high 2.73 0.94 Eat pecans 1 = Very often/often/sometimes 0.76 0.43 Look for COO on food labels 1 = Always or often 0.36 0.48 Aware of mandatory pecan COO label 1 = Yes 0.30 0.46

The primary feature of each session is an experimental auction to measure consumers’

willingness to pay for pecan products, employed in order elicit a direct estimate of WTP,

including the effect of different labeling options.

3 We used the same method as Klain et al. (2014), which involved shortening the CETSCALE found in Shimp and Sharma (Shimp and Sharma 1987) from a set of 17 Likert-scaled question down to three questions. Their test of the original measure led to choosing the three that had the highest factor loadings in the general ethnocentrism scale, leading to development of a 3-question based average. This average was calculated based on the respondent’s answer to three questions where each individuals’ responses ranged from “strongly agree” to “strongly disagree” on a five-point Likert-scale. The exact language in the three questions as quoted from Klain were: “Americans should not buy foreign products, because this hurts American business and causes unemployment,” “It is not right to purchase foreign products because it puts Americans out of jobs,” and “A real American should always buy American made products.”

68

Experiment Steps

Figure 3.1 is a flowchart outlining the experiment steps for participants. For Step 1, upon arrival,

the participants check in, read and sign a consent form, receive their $50 participation fee, along

with a packet containing all materials required to participate in the experiment. See Appendix A

for a copy of the participant materials packet. The information is identified only with an ID

number to preserve anonymity and all responses turned in by participants contain only the ID

number, with no individually identifying information. The packet includes the pre-auction

questionnaire mentioned above and page-by-page instructions to follow during the one-hour

session. Once seated in the lab, participants complete the basic demographics portion of the

questionnaire before proceeding to the next step, led by the monitor.

For Step 2, the session monitor informs participants that they will be engaging in an

auction of common food products and asked to refrain from communicating with others during

the process since the bid submissions are private They receive an explanation of the type of

auction to be conducted, that it is likely different than other types of auctions in which they have

previously participated, they will be bidding on six product options, with only one that counts (is

“binding”), and auction winners make an actual purchase of the one product selected. Further

explanation is given to participants, with instructions and written examples, quiz and practice

rounds, and precisely how the winner is chosen from the submitted bids. The auction method is

the random nth-price auction (Lusk and Shogren 2007, Rousu et al. 2007, Colson, Huffman, and

Rousu 2011)

In a random nth price auction, k bidders – for example k = 15 – confidentially submit

their non-negative bid for the pecans under consideration on the bid sheet provided in their

information packet. Once submitted, the session monitor places bids in order ranked from

69

highest to lowest. Next, the monitor draws the “random n” from a uniform distribution of

positive integers between 2 and k, for example n=4. This drawn “nth-price” becomes the nth

highest bid and n-1 individuals submitting bids higher than this win the auction, with each

paying the nth ranked price for the object (Colson 2009), not their own submitted bid price. In

our illustrative example, that would mean that the three highest bidders are the auction winners

(n (4)-1=3) and all three pay the 4th highest price.4

The random nth price auction elicitation device was chosen because of its prevalent use

in food values research (McFadden and Huffman 2017) and because it combines the advantages

of both stated and revealed preference, is demand revealing (Lusk and Shogren 2007) and is

incentive compatible because each participant has an incentive to bid their own private and true

value of the product. To better understand the demand revealing nature of the method, we build

upon the intuition of Vickrey’s classic second-price auction (Vickrey 1961), where 𝑘𝑘 bidders

have independent private values 𝜐𝜐𝑘𝑘 and submit sealed bids 𝑏𝑏𝑘𝑘 with payoffs 𝜋𝜋𝑘𝑘. Let 𝛽𝛽 denote the

nth-highest bid from the random integer, ∈ {2, 3,} and nature selects 𝜐𝜐𝑘𝑘 and 𝑛𝑛.

𝜋𝜋 𝑘𝑘 = �𝜐𝜐𝑘𝑘 − 𝛽𝛽 𝑖𝑖𝑖𝑖 𝑏𝑏𝑘𝑘 > 𝛽𝛽0 𝑖𝑖𝑖𝑖 𝑏𝑏𝑘𝑘 ≤ 𝛽𝛽

This payoff structure demonstrates how the dominant strategy for each bidder is to bid their own

private value for the product, or 𝑏𝑏𝑘𝑘 = 𝑣𝑣𝑘𝑘.

As opposed to other auctions, in which there is only one auction winner, the random nth

price auction provides an opportunity for n-1 bidders to win the auction and purchase the

product. Further, as opposed to other demand-revealing auctions (e.g. Vickrey or Becker,

DeGroot, and Marschak (1964)), this method better engages the off-margin bidders with low

4 As mentioned in Rousu et al. (2007), if there are ties present in the ranking, it can result in the number of winners being greater than n-1. This allows the advantage of these bidders receiving product at a price lower than their bid.

70

values for the product being auctioned (Rousu et al. 2007) because they still have positive

probability of being an auction winner. McFadden and Huffman (2017) find evidence of more

sincere bidding from low-value individuals under this method. Another advantageous property

is use of a random endogenously determined market-clearing price (Shogren et al. 2001). The

attributes of this method provide a good fit for the study objectives.

Step 3 begins with the monitor placing fictional nth price auction bids on a whiteboard

visible to all in the front of the room, demonstrating the ranking process and how auction

winners are determined. A quiz follows with answers to ensure a thorough grasp of how the

auction proceeds. Participants then engage in two practice nth price auction rounds, first using

black bean soup followed by chicken broth in order to gain experience with familiar products in

a hypothetical practice round.

71

Figure 3.1: Details of experiment steps

This step introduces practice for the choice of bidding between products in separate

opaque boxes containing the different label treatments as in the actual experiment. Specifically,

participants are asked to list their bids (hypothetically, for practice rounds) for a 15 oz. can of

black bean soup (or chicken broth in practice round 2) from inside containers under three

different label options. Participants are not allowed to view the product within the containers,

instead shown details about what information is present on the label of the product within each

container, that are identical in every other way.

72

This practice round introduces the concept of risk by including a label option of a 50/50

probability of an organic product. See Figure 3.2 for example of practice round bid sheet

containing all three options as viewed by bidders. For both practice rounds, monitors collect the

bids, and rank in order from highest to lowest bid. From there, a coin toss determines which of

the two practice rounds will be considered the binding round, followed by a random draw to

select the nth price. Following the practice rounds, monitors answer any remaining questions

about the auction procedures.

73

Figure 3.2: Example of practice round for auction

74

Following practice rounds, Step 4 begins the experimental auction portion of the study.

Participants are shown a bag of pecans without labels for familiarity with the product, but not

allowed to look inside of six containers, one for each label treatment option on which they bid.

Before bidding commences, monitors place the six containers on a table, displaying them side-

by-side, each clearly displaying product label information but with actual pecan bags inside and

not visible to participants. The label options include no information (all options contain Pecan

Halves, 8 oz.), Product of the United States, Product of Mexico, Product of Mexico and/or the

United States, Contains 50% Pecans from Mexico & 50% Pecans from the United States, and

50% Probability Product of Mexico & 50% Probability Product of the United States. No

explanation beyond what is on the bid sheets and label wording is given to bidders.

The concept of uncertainty introduced in the practice round is expanded in the actual

experiment to explore the influence of risk and/or ambiguity in the willingness to pay for the

shelled pecan halves under the different labeling options. The “risk” present for the auction bids

consists of some probability of determining the country of origin of the product, while

“ambiguity” for our purposes indicates that those probabilities are not known. For example, in

the case of the blank label, where only the product name and package weight are revealed, both

risk and ambiguity are present since there is some probability of knowing the source country and

the probability is not known because the label does not specify. As a second example, in the case

of the 50/50 Probability Product of Mexico or United States label, there is risk for this option since

there is a probability of determining the origin, a 50% probability of product from Mexico

(single origin) and 50% probability from U.S. (single origin), but there is no ambiguity present as

the probabilities are known. See Table 3.2 for the available label treatment options and full

descriptions of the risk and ambiguity present for the bidder.

75

For each treatment, we elicit the highest (non-negative) price the participant is willing to

pay for pecans with the specific labeled option. These bids are submitted simultaneously in one

round of bidding and collected by the monitor. Figure 3.3 is the bidding and instruction sheet for

the auction.

Table 3.2: Label treatment options Label Option Risk Present?a Ambiguity Present?b

A Blank Yesc Yesc B Product of the United States No No C Product of Mexico No No D Product of Mexico and/or the United States Yes Yes E 50/50 Product of Mexico and United States No No F 50/50 Probability Product of Mexico or United States Yes No

Notes: a “Risk” denotes that there is some probability, known or unknown, of determining the origin b “Ambiguity” denotes that the probabilities are not known. c There is only risk/ambiguity if consumers actually consider the possible origin of the pecans in submitting their bids

For Step 5, as demonstrated in the practice rounds, bids are sorted and ranked from

highest to lowest and a random draw selects which label option will be binding as well as the

random n, then the winning bid ID numbers are posted on the whiteboard, along with the nth

price that will be paid by the n-1 winners.

Step 6 concludes the experiment to elicit consumer valuation for the pecan label options

and all participants answer the post-auction questionnaire and participate in the decision tasks

used to measure their risk and ambiguity attitudes. Upon completion, auction winners exchange

money for the bag of pecans from the binding label treatment, while the non-winners exit the

room. All options presented in label treatments were available for auction winners for purchase,

following Colson et al. (2016) and Rousu et al. (2015) to ensure that no deceptive practices were

employed in the experiment.

76

Figure 3.3: Example of auction label treatment round bid sheet

77

Methods

In this experiment, we elicit participant bids on 8 oz. bags of pecans in one auction round under

six different label treatment options. Multiple regression analysis is employed to address any

potential confounding factors, such as the participant’s unique socioeconomic characteristics.

The method theoretically holds constant the participants taste between pecans in the six

containers, where the dependent variable is the highest (non-negative) dollar value that they are

willing to pay for the pecans with the given characteristics. Consistent with this approach, the

participant’s inverse demand equation for pecans is (Rousu et al. 2007):

𝑃𝑃𝑖𝑖𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 𝑇𝑇𝑇𝑇𝐿𝐿𝐿𝐿𝑇𝑇𝑇𝑇𝐿𝐿𝑇𝑇𝑇𝑇 𝑂𝑂𝑂𝑂𝑇𝑇𝑖𝑖𝑂𝑂𝑇𝑇 = 𝛽𝛽1 + 𝛽𝛽2𝑋𝑋𝑖𝑖2 + 𝜀𝜀𝑖𝑖,

where 𝑃𝑃𝑖𝑖is the price bid by participant i for the particular label treatment option, 𝛽𝛽1 is the

intercept, 𝑋𝑋𝑖𝑖2 is a vector of explanatory variables for each individual, 𝛽𝛽2 is a vector of

coefficients, and 𝜀𝜀𝑖𝑖 is the error term. Under this specification, the experiment design results in a

bid for willingness to pay, in terms of dollars and cents for selecting pecans with a specific label

treatment option.

The perspective of risk and willingness to pay for food products has been largely

addressed in terms of food safety (for example, see Loureiro and Umberger (2007), Lim et al.

(2013), Lim et al. (2014), Lewis and Grebitus (2016)), typically exploring risk in the context of

potential hazards involved in purchasing a meat product that may be unsafe or perceived as

unsafe for consumption. For this study, we look at how risk and ambiguity in knowing the

country of origin for auction bidding may influence consumers’ willingness to pay varying label

treatment options. The experimental method used to jointly elicit participant risk and ambiguity

attitudes for this study is based on the approach introduced by (Gneezy, Imas, and List 2015),

built upon the framework of (Holt and Laury 2002) for risk aversion and Ellsberg (1961) urn

78

techniques for ambiguity. Gneezy, Imas, and List (2015) stresses the importance of jointly

estimating the risk and ambiguity attitudes to avoid overestimating ambiguity aversion by

assuming a level of risk aversion rather than simultaneously obtaining the measures. As outlined

for this paper-and-pencil instrument, the joint elicitation requires the individual to make a series

of decisions under both subjective and objective uncertainty using the double multiple price list

(DMPL) format, then using that data to estimate the risk aversion coefficient 𝑟𝑟 and ambiguity

aversion measure 𝛼𝛼. Specifically designed to be simple and easy to use in the field or laboratory,

the approach makes minimal assumptions and restrictions.

Procedurally, this method involves asking individuals to make decisions in two separate

“tasks” on forms (a “decision sheet”) following the post-auction questionnaire, each a multiple

price list (MPL). The first task, Task A, includes a set of ten decisions made over choices where

the probabilities are objective. The second task, Task B, is a set of twenty decisions between

choices where some probabilities are known and others ambiguous. Each decision sheet

preceded by instructions read by participants and also with an explanation by the session

monitor. See Appendix A for examples of the instructions and decision sheet for each task. The

focus of the responses is the point at which the respondent switches from Option A in the left

column or the gamble Option B in the right. The “switch point” selection is the data from which

the risk attitude for Task A and ambiguity attitude is calculated for Task B. These experimental

tasks are designed to be incentive compatible, with respondents receiving payment according to

the randomly selected task and decision, following payment if auction winner. Like the Gneezy

design, approximately one half of the research session respondents receive real money from their

decision task while the others chose from hypothetical gambles. In the method development,

they found no statistical difference between real or hypothetical incentives, nor did we (see table

79

3.3). Our refinements to the approach include paying exact dollar amounts for the real gambles

as opposed to tokens as in the Gneezy experiments.

Table 3.3: Aggregate risk and ambiguity attitudes under real and hypothetical incentives (N=129) Risk Coefficient 𝑟𝑟 Ambiguity Attitude 𝛼𝛼 Hypothetical Incentives Joint Estimate 0.475 0.555 (0.285) (0.457) Real Incentives Joint Estimate 0.479 0.601 (0.285) (0.357) Difference -0.004 -0.046 (0.051) (0.074)

Standard deviation for joint estimate, standard errors for difference in parentheses.

Data Analysis and Results

Table 3.4 presents summary statistics of the auction bid prices, with unconditional tests

of differences between bids across the six label treatment options, exploring whether single

source country of origin and mixed country of origin in the presence of various risk and

ambiguity scenarios influence consumer valuation for 8 oz. bags of shelled pecan halves.

Despite identical product contained within, the bidders’ valuation for the products range from

$3.40 to $4.54 per 8 oz. bag.5 At $4.54, the highest average value (label option F) is for the

unambiguous Product of the United States, topping the next highest option of $3.74 (label

treatment A) for the blank label containing only the product name and weight with no country of

origin data. At the other end of the spectrum, the lowest in preference ordering is the average

bid is for the single origin Product of Mexico label ($3.40 – label treatment C), though the value

is similar to the average bid for known-probability risk (50/50) of purchasing a single country

source product from either the United States or Mexico ($3.44 – label treatment F). The average

5 The approximate retail value for the product in grocery stores is $5.99.

80

bid for label treatment D with non-specified origin information at $3.59 is the same on average

as the clearly defined 50/50 mixed origin product from U.S. and Mexico with no risk or

ambiguity for a determination of origin (treatment E). Similar to the Klain et al. (2014) findings

for consumer preferences for meat origin, any label information that includes or may include

Mexico origin are the least preferred of all available treatment options.

Table 3.4: Summary statistics of auction bid prices (N=129) Summary Statistics by Label Treatment Options* Treatment Label Information Mean Std. Deviation

A Blank $3.74 $2.04 B Product of U.S. $4.54 $2.32 C Product of Mexico $3.40 $2.05 D Product of Mexico and/or the United States $3.59 $2.12 E 50/50 Product of Mexico and United States $3.59 $2.11 F 50/50 Probability Product of Mexico or United States $3.44 $2.04

.

In table 3.5, the difference in bids between all label treatments and statistical tests of

differences from zero is presented, exploring how varying country of origin options affect the

average willingness to pay among the bidders. Overall, there are consistent and significant

differences in bid prices between the label information options. The largest mean bid difference

($1.15) is between Product of the United States over the Product of Mexico pecans. On average,

consumers bid between significantly more (between $0.95 and $1.15) for the U.S. product over

all other label treatment options. There is no significant difference between treatment D and E,

nor D and C. With the exception of the guaranteed single origin U.S. product, the differences

between the blank label (treatment A) and other options are positive, meaning that the average

bid for product from Mexico or any mixture of origins is less preferred than knowing nothing but

product name and weight. Though seemingly paradoxical that reliable, additional information

about product origin results in lower average bids, it suggests several possibilities. It could be

81

that consumers are confused as the labels become more complex (Kuchler et al. 2017), that

consumers may infer or assume certain characteristics about the product despite the blank label,

or that the country of origin may not have been part of their preference profile or purchase

decision process prior to it being pointed out on the label.

Table 3.5: Mean bid difference between all pairs of label treatments Mean bid difference between all label information options (N = 129) Treatment Label Information B C D E F

A Blank -$0.80*** $0.34*** $0.15* $0.15* $0.30*** (0.08) (0.08) (0.08) (0.08) (0.09)

B Product of the United States $1.15*** $0.95*** $0.96*** $1.11*** (0.10) (0.09) (0.10) (0.10)

C Product of Mexico -$0.19*** -$0.19*** -$0.04 (0.05) (0.06) (0.07)

D Product of Mexico and/or the United States $0.00 $0.15** (0.07) (0.07)

E 50/50 Product of Mexico and United States $0.15*** (0.06)

F 50/50 Probability Product of Mexico or U.S. aStandard errors in parentheses. Difference is calculated using the bid on vertical columns minus bid on horizontal top column. Statistical significance * p<0.10, ** p<0.05, *** p<0.01

We find that, relatively, consumers are significantly willing to pay less to obtain a single

origin product under the known probability of risk (treatment F) than for any other single or

mixed origin product offered, besides the known origin product from Mexico. This result could

also seem contradictory, for example, that bidders on average would be willing to pay $0.15

more for the ambiguous risk of determining the country of origin (treatment D – “and/or”) or the

50/50 product origin mix under no risk or ambiguity (treatment E). In the case of the higher

average bids for D and E, the pecans are either potentially or known mixed origin, so bidders

appear to prefer a mix of pecans from both countries over the 50% chance of receiving a single

origin product that is less preferred, as the pecans from Mexico appear to be.

82

To further investigate preferences for the different label options and the influence of risk

and ambiguity attitudes on consumer willingness to pay, we expand our definitions and clarify

relevant bid dissimilarities in Table 3.6, presenting the mean, hypothesized sign, and statistical

tests of differences from zero, along with a definition of each with respect to hypotheses.

Consistently, the largest premium is for the U.S. single origin product compared to the product

known to be from Mexico. The unambiguously certain single country of origin premium (U.S.

vs. Mexico, $1.15) highlights the premise of well-defined preferences and certainty over

uncertainty.

From the perspective of contrasting single and mixed origin bids, we make comparisons

between the weighted-average bids for a single origin pecan product with the mixed origin

treatment options. From this angle, results reveal the largest discount in willingness to pay for

mixed vs. single origin (-$0.54) when the risk is both explicit on the label and the probability is

known for the mixed origin pecans. Similarly, bidders reveal a discount (-$0.38) for mixed origin

pecans over single origin under both risk and ambiguity or if the pecans are a known 50/50 mix

from both U.S. and Mexico.

83

Table 3.6: Relevant Bid Differences and Definitions (note changes: simpler version than we had last discussed) Bid Differences Expected Sign Mean Definition

B-C >0 $1.15 *** Premium for single origin product of United States vs. Mexico

D-A ≷ 0 -$0.15 * Discount/Premium for explicita vs. implicita origin ambiguity

D-F ≷ 0 $0.15 ** Discount/Premium for explicit origin risk vs. known-risk single origin

A-F ≷ 0 $0.30 *** Discount/Premium for implicit origin risk vs. known-risk single origin

D-E <0 $0.00 Discount for explicit origin risk vs. explicit mixed origin

D-(0.5B + 0.5C) <0 -$0.38 *** Discount for explicit origin risk vs. single origin product

E-(0.5B + 0.5C) <0 -$0.38 *** Discount for mixed vs single origin product

F-(0.5B + 0.5C) <0 -$0.54 *** Discount for risk a Explicit origin ambiguity refers to that which is brought to the attention of the consumer as a result of being on the label. Implicit indicates that the same ambiguity is present for the product but it is not noted on the label. Statistical significance * p<0.10, ** p<0.05, *** p<0.01

Beyond the unconditional analysis of the average bids, we look at the potential drivers of

the consumer willingness to pay for single and mixed origin pecans across different label

contexts. For this purpose, we present a series of regressions looking at the influence of socio-

demographic, shopping preferences, consumer ethnocentrism, and risk and ambiguity attitudes

on bid prices. For each label treatment option and for the bid differences, we estimate standard

linear regression models (OLS). We find few variables statistically significant with consistency

across the models and treatments. Noteworthy results in the estimates for each label treatment

(table 3.7) is the positive effect of bidder certainty (ranging from $0.21 to $0.29) across all six

options and from female bidders ($0.69 to $0.93) for all but treatment F. In the regressions for

relevant bid differences (table 3.8), we find limited evidence of a premium for consumers who

look for country of origin on food labels between the single origin products, but a discount

among them in comparison between single and mixed origin products. We do not find any

consistent support for our measure of consumer ethnocentrism influencing values nor for any of

our risk and ambiguity attitudes, despite the clear preference away from single origin products

from Mexico. Overall, our regression analysis offers little meaningful understanding into the

84

factors influencing consumer preferences for single or mixed country of origin information in

pecan purchases, even in the presence of risk and ambiguity for these purchases.

Conclusion

Providing attributes that consumers desire and have confidence in can prove challenging for food

products, particularly with respect to successfully relaying information on the label. With a

multi-attribute product or with multi-national sourcing, these issues grow more complicated in

not only understanding and addressing well-defined consumer preferences, but also complying

with regulations concerning specifics. Country of origin labeling requirements provide an

example of these complexities, with producers facing obstacles of potential product

commingling, seasonality, and different rules depending on whether product is sourced from a

single country or from multiple. Further, if consumers are uncertain about product attributes or

origin, it can threaten the price premium needed to maintain profit margins.

In this study, we conduct a series of experimental auctions to elicit the direct willingness

to pay for pecans under labeling scenarios that include single and mixed country of origin

products, both with and without explicit risk and/or ambiguity of knowing the origin. On

average, consumers value the single origin product from the U.S. with certainty over all other

options, with the relatively least preferred being the product with a known risk of potentially

obtaining the single origin product from Mexico. Despite the implicit, yet still present, risk of

not knowing the country of origin where the label is blank, consumers appeared to have either

not understood or not considered the risk and ambiguity of the choices unless pointed out to them

by its occurrence explicitly on the label. The overall findings indicate a preference ordering of

single origin product over mixed origin, regardless of whether the risk of determining the origin

is known or unknown. Because consumers show a clear preference of willingness to pay more

85

for unambiguous single origin product, it follows that single source producers might favor the

mandatory COO labeling as currently exists in order to capture those premiums. However, those

with product from mixed or potentially mixed origins might favor a voluntary COO policy in

order to avoid lower consumer valuations. Consumers are indifferent between product that

specifies what fraction comes from each country (50/50 U.S. and Mexico) or simply stating

“and/or” (U.S. and/or Mexico) on the label, which is not currently allowed under COO

regulation.

Though a clear pattern of preference ordering appears in the unconditional analysis of the

bids, conditional analysis reveals no distinct influence of the socio-demographic, shopping

patterns, or risk and ambiguity attitudes in these consumer valuations. This finding could reflect

the signaling nature of the country of origin attribute, that it is more than just a reduction of

asymmetric information to the consumer, but incorporates multiple dimensions of the consumer

preferences such as a quality marker or a way to support one’s native country. As a result, these

differences across the average consumer may not fall into expected categories.

For future research efforts, some variation in experiment design might prove fruitful in

untangling the details of the primary influences of consumer willingness to pay in the presence

of both risk and ambiguity for particular product attributes. Some areas to explore include using

a different commodity of focus, comparing two products that have different inherent risk

profiles, or to examine other attributes in conjunction with COO such as a known brand name to

detail consumer willingness to pay.

86

Table 3.7: Regression estimates for bids Treatment A Treatment B Treatment C Treatment D Treatment E Treatment F Risk Aversion (CRRA coeff.) -0.245 -0.037 -0.103 -0.136 -0.134 -0.117 (0.242) (0.251) (0.249) (0.248) (0.245) (0.237) Alpha (Ambiguity attitude coeff.) 0.046 0.338 0.184 0.169 0.187 0.142 (0.441) (0.531) (0.509) (0.491) (0.498) (0.487) Bid Certainty (true maximum WTP) 0.227*** 0.291*** 0.247*** 0.267*** 0.210** 0.257*** (0.083) (0.090) (0.076) (0.080) (0.096) (0.088) Female 0.898** 0.930** 0.686* 0.739* 0.803** 0.603 (0.352) (0.391) (0.365) (0.383) (0.387) (0.388) Age 0.018 0.023 0.010 0.015 0.004 0.009 (0.014) (0.015) (0.014) (0.014) (0.014) (0.014) Income -0.116 -0.063 0.109 0.067 0.130 0.060 (0.336) (0.377) (0.381) (0.384) (0.382) (0.376) Education 0.029 0.005 0.022 0.014 0.004 0.001 (0.076) (0.078) (0.076) (0.076) (0.076) (0.067) Household size -0.050 -0.032 0.029 0.036 0.033 0.013 (0.138) (0.138) (0.134) (0.138) (0.136) (0.134) Consumer Ethnocentrism -0.070 0.024 -0.062 -0.008 -0.031 -0.100 (0.242) (0.246) (0.238) (0.247) (0.239) (0.217) Eat pecans -0.172 -0.111 -0.449 -0.468 -0.327 -0.472 (0.412) (0.460) (0.469) (0.461) (0.447) (0.440) Look for COO on food labels 0.261 0.682 0.235 0.076 0.143 0.322 (0.436) (0.440) (0.468) (0.460) (0.471) (0.425) Aware of mandatory COO on pecans 0.632 0.501 0.160 0.252 0.480 0.194 (0.392) (0.447) (0.419) (0.429) (0.422) (0.402) Constant 0.367 0.229 0.496 0.298 1.021 1.029 (1.381) (1.349) (1.370) (1.336) (1.511) (1.355) N 123 123 123 123 123 123 adj. R-sq 0.131 0.173 0.061 0.071 0.048 0.061 F 3.366*** 3.725*** 1.895** 2.137** 1.778* 1.862** Standard errors in parentheses * p<0.10, ** p<0.05, *** p<0.01

87

Table 3.8: Regression estimates for bid differences B-C D-A D-F A-F D-E D-

(.5B+.5C) E-

(.5B+.5C) F-

(.5B+.5C) Risk Aversion (CRRA coeff.) 0.066 0.109 -0.020 -0.129 -0.002 -0.066 -0.064 -0.047 (0.114) (0.099) (0.084) (0.102) (0.082) (0.059) (0.077) (0.079) Alpha (Ambiguity attitude coeff.) 0.154 0.123 0.027 -0.096 -0.018 -0.092 -0.074 -0.119 (0.279) (0.243) (0.208) (0.252) (0.202) (0.144) (0.189) (0.195) Bid Certainty (true maxWTP) 0.045 0.040 0.011 -0.029 0.058* -0.002 -0.059** -0.013 (0.043) (0.038) (0.032) (0.039) (0.031) (0.022) (0.029) (0.030) Female 0.244 -0.159 0.137 0.295 -0.064 -0.069 -0.005 -0.205 (0.205) (0.179) (0.153) (0.185) (0.149) (0.106) (0.139) (0.143) Age 0.013* -0.002 0.006 0.008 0.011** -0.001 -0.013** -0.007 (0.007) (0.006) (0.006) (0.007) (0.005) (0.004) (0.005) (0.005) Income -0.172 0.183 0.008 -0.176 -0.063 0.044 0.107 0.037 (0.220) (0.191) (0.164) (0.198) (0.159) (0.114) (0.149) (0.153) Education -0.018 -0.016 0.012 0.028 0.010 0.000 -0.009 -0.012 (0.044) (0.038) (0.033) (0.039) (0.032) (0.023) (0.030) (0.031) Household size -0.061 0.087 0.023 -0.064 0.003 0.038 0.035 0.015 (0.073) (0.064) (0.055) (0.066) (0.053) (0.038) (0.050) (0.051) Consumer Ethnocentrism 0.086 0.062 0.092 0.030 0.024 0.011 -0.013 -0.081 (0.126) (0.110) (0.094) (0.114) (0.091) (0.065) (0.085) (0.088) Eat pecans 0.339 -0.296 0.004 0.300 -0.141 -0.188 -0.046 -0.192 (0.250) (0.217) (0.186) (0.225) (0.181) (0.129) (0.169) (0.174) Look for COO on food labels 0.447* -0.185 -0.245 -0.061 -0.067 -0.382*** -0.316* -0.137 (0.240) (0.208) (0.178) (0.216) (0.173) (0.124) (0.162) (0.167) Aware of mandatory COO on pecans 0.342 -0.380* 0.058 0.438* -0.228 -0.078 0.149 -0.136 (0.247) (0.215) (0.184) (0.223) (0.179) (0.128) (0.167) (0.172) Constant -0.267 -0.069 -0.731 -0.662 -0.723 -0.064 0.659 0.666 (0.817) (0.711) (0.608) (0.737) (0.592) (0.423) (0.552) (0.570) N 123 123 123 123 123 123 123 123 adj. R-sq 0.123 0.043 -0.048 0.053 0.008 0.099 0.099 0.048 F 2.429*** 1.452 0.534 1.568 1.081 2.120** 2.120** 1.509


88

References

Becker, Gordon M, Morris H Degroot, and Jacob Marschak. 1964. "Measuring Utility by a Single‐Response Sequential Method." Systems Research and Behavioral Science 9 (3):226-232.

Colson, Gregory. 2009. "Improving Nutrient Content through Genetic Modification: Evidence from Experimental Auctions on Consumer Acceptance and Willingness to Pay for Intragenic Foods." Iowa State University (10801).

Colson, Gregory, Jay R. Corrigan, Carola Grebitus, Maria L. Loureiro, and Matthew C. Rousu. 2016. "Which Deceptive Practices, If Any, Should Be Allowed in Experimental Economics Research? Results from Surveys of Applied Experimental Economists and Students." American Journal of Agricultural Economics 98 (2):610-621.

Colson, Gregory J, Wallace E Huffman, and Matthew C Rousu. 2011. "Improving the Nutrient Content of Food through Genetic Modification: Evidence from Experimental Auctions on Consumer Acceptance." Journal of Agricultural and Resource Economics:343-364.

Devuyst, Ea, Jl Lusk, and Ma Devuyst. 2014. "The USDA Quality Grades May Mislead Consumers." Journal of Animal Science 92 (7):3142-3148.

Ehmke, Mariah Tanner. 2006. "International Differences in Consumer Preferences for Food Country-of-Origin: A Meta-Analysis." American Agricultural Economics Association Meeting, Long Beach CA.

Ellsberg, Daniel. 1961. "Risk, Ambiguity, and the Savage Axioms." The Quarterly Journal of Economics:643-669.

Florkowski, Wojciech J, Zhikang You, and Chung L Huang. 1999. "Consumer Selection of Retail Outlets in Buying Pecans." Journal of Food Distribution Research 30:34-43.

Gao, Zhifeng, Ted C Schroeder, and Xiaohua Yu. 2010. "Consumer Willingness to Pay for Cue Attribute: The Value Beyond Its Own." Journal of International Food & Agribusiness Marketing 22 (1-2):108-124.

Gneezy, Uri, Alex Imas, and John List. 2015. Estimating Individual Ambiguity Aversion: A Simple Approach. National Bureau of Economic Research.

89

Gold, Michael, Mihaela M Cernusca, and Larry Godsey. 2004. "Consumer Preferences for Chestnuts, Eastern Black Walnuts, and Pecans." HortTechnology 14 (4):583-589.

Greene, Joel L. 2016. "Country-of-Origin Labeling for Foods and the WTO Trade Dispute on Meat Labeling."

Hayes, Dermot J, and Sergio H Lence. 2015. "A New Brand of Agriculture? Farmer-Owned Brands Reward Innovation." Iowa Ag Review 8 (4):2.

Holt, Charles A, and Susan K Laury. 2002. "Risk Aversion and Incentive Effects." American Economic Review 92 (5):1644-1655.

Klain, Tyler J., Jayson L. Lusk, Glynn T. Tonsor, and Ted C. Schroeder. 2014. "An Experimental Approach to Valuing Information." Agricultural Economics 45 (5):635-648.

Kuchler, Fred, Catherine Greene, Maria Bowman, Kandice K. Marshall, John Bovay, and Lori Lynch. 2017. Beyond Nutrition and Organic Labels-30 Years of Experience with Intervening in Food Labels, ERR-239,. edited by U.S. Department of Agriculture Economic Research Service.

Lewis, Karen E, and Carola Grebitus. 2016. "Why Us Consumers Support Country of Origin Labeling: Examining the Impact of Ethnocentrism and Food Safety." Journal of International Food & Agribusiness Marketing 28 (3):254-270.

Lillywhite, Jay M, Jennifer E Simonsen, and Richard J Heerema. 2014. "US Consumer Purchases and Nutritional Knowledge of Pecans." HortTechnology 24 (2):222-230.

Lim, Kar H, Wuyang Hu, Leigh J Maynard, and Ellen Goddard. 2014. "A Taste for Safer Beef? How Much Does Consumers’ Perceived Risk Influence Willingness to Pay for Country‐of‐Origin Labeled Beef." Agribusiness 30 (1):17-30.

Lim, Kar H., Wuyang Hu, Leigh J. Maynard, and Ellen Goddard. 2013. "U.S. Consumers' Preference and Willingness to Pay for Country-of-Origin-Labeled Beef Steak and Food Safety Enhancements." Canadian Journal of Agricultural Economics 61 (1):93-118.

Lombardini, Leonardo, Tina M Waliczek, and Jayne M Zajicek. 2008. "Consumer Knowledge of Nutritional Attributes of Pecans and Factors Affecting Purchasing Behavior." HortTechnology 18 (3):481-488.

90

Loureiro, Maria L., and Wendy J. Umberger. 2003. "Estimating Consumer Willingness to Pay for Country-of-Origin Labeling." Journal of Agricultural and Resource Economics 28 (2):287-301.

Loureiro, Maria L., and Wendy J. Umberger. 2005. "Assessing Consumer Preferences for Country-of-Origin Labeling." Journal of Agricultural and Applied Economics 37 (1):49-63.

Loureiro, Maria L., and Wendy J. Umberger. 2007. "A Choice Experiment Model for Beef: What Us Consumer Responses Tell Us About Relative Preferences for Food Safety, Country-of-Origin Labeling and Traceability." Food Policy 32 (4):496-514.

Lusk, Jayson L, and Jason F Shogren. 2007. Experimental Auctions: Methods and Applications in Economic and Marketing Research: Cambridge University Press.

Mabiso, Athur, James Sterns, Lisa House, and Allen Wysocki. 2005. "Estimating Consumers’ Willingness-to-Pay for Country-of-Origin Labels in Fresh Apples and Tomatoes: A Double-Hurdle Probit Analysis of American Data Using Factor Scores." American Agricultural Economics Association Annual Meeting, Providence, Rhode Island.

McFadden, Jonathan R, and Wallace E Huffman. 2017. "Willingness-to-Pay for Natural, Organic, and Conventional Foods: The Effects of Information and Meaningful Labels." Food Policy 68:214-232.

Noussair, Charles, Stephane Robin, and Bernard Ruffieux. 2002. "Do Consumers Not Care About Biotech Foods or Do They Just Not Read the Labels?" Economics letters 75 (1):47-53.

Palma, Marco A, Alba J Collart, and Christopher J Chammoun. 2015. "Information Asymmetry in Consumer Perceptions of Quality‐Differentiated Food Products." Journal of Consumer Affairs 49 (3):596-612.

Park, Timothy A, and Wojciech J Florkowski. 1999. "Demand and Quality Uncertainty in Pecan Purchasing Decisions." Journal of Agricultural and Applied Economics 31 (1):29-39.

Rousu, Matthew C, Gregory Colson, Jay R Corrigan, Carola Grebitus, and Maria L Loureiro. 2015. "Deception in Experiments: Towards Guidelines on Use in Applied Economics Research." Applied Economic Perspectives and Policy 37 (3):524-536.

91

Rousu, Matthew, Wallace E. Huffman, Jason F. Shogren, and Abebayehu Tegene. 2007. "Effects and Value of Verifiable Information in a Controversial Market: Evidence from Lab Auctions of Genetically Modified Food." Economic Inquiry 45 (3):409-432.

Shimp, Terence A., and Subhash Sharma. 1987. "Consumer Ethnocentrism: Construction and Validation of the Cetscale." Journal of Marketing Research 24 (3):280-289. doi: 10.2307/3151638.

Shogren, Jason F, Michael Margolis, Cannon Koo, and John A List. 2001. "A Random Nth-Price Auction." Journal of Economic Behavior & Organization 46 (4):409-421.

Tonsor, Glynn T, Ted C Schroeder, and Jayson L Lusk. 2013. "Consumer Valuation of Alternative Meat Origin Labels." Journal of Agricultural Economics 64 (3):676-692.

USDA, Agricultural Marketing Service. 2017. "Country of Origin Labeling Fiscal Year 2016 Retail Compliance Data." accessed March 13, 2018. https://www.ams.usda.gov/sites/default/files/media/2016RetailComplianceData.pdf.

USDA, Agricultural Marketing Service. n.d. "Marketing Orders & Agreements." accessed June 26, 2017. https://www.ams.usda.gov/rules-regulations/moa.

Vickrey, William. 1961. "Counterspeculation, Auctions, and Competitive Sealed Tenders." The Journal of Finance 16 (1):8-37.

92

CHAPTER 4: GENERAL CONCLUSION

Pecans are a high-value component of U.S. agriculture, yet little information about

particular consumer preferences along with the corresponding willingness to pay (WTP) for this

native North American tree nut can be found. Because pecans are considered a “covered

commodity” subject to mandatory country of origin (COO) labeling requirements by retailers

under United States Department of Agriculture guidelines, questions related to this requirement

and consumer valuations emerge alongside the others. Because product labels convey vital

information from producers and manufacturers to consumers, they are key in investigating

consumer valuations. Clear and effective labels not only allow consumers to align their

preferences with available options, but hold potential economic gains when product information

garners a price premium for desired attributes.

Employing choice experiments and random nth price experimental auctions in a series of

research sessions with adult consumers in two Southeastern U.S. cities, this study assesses how

consumers respond to product attributes and label details in different information contexts for

purchases of shelled pecan halves. From the data collected in the sessions, we address the

following: 1) consumers’ willingness to pay for select initiatives proposed in conjunction with

the recent Federal Marketing Order (FMO) regarding pecan attributes, 2) direct elicitation of the

consumers’ value of obtaining information about attributes under alternative labeling scenarios,

and 3) the influence of consumer risk and ambiguity attitudes on consumer preferences for single

or mixed country of origin products.

93

In Chapter 1, we look at how long standing issues in the U.S. pecan industry ultimately

led to the 2016 approval of a Federal Marketing Order (FMO), which underscores the

importance of understanding consumer valuation as stakeholders pursue avenues to address

industry-wide concerns that include supply, prices, market disruption, and lack of uniform

quality, container, and packaging standards. Yet, there are no guarantees that the passing of the

FMO or any subsequent actions that may be implemented will bring about the anticipated

increased demand, additional profits, or better prices for pecan producers. Our findings reveal

that the most consistent overall attribute for which consumers are willing to pay is for pecans

sourced from the U.S., the labeling of which is already a requirement by COO regulation. In

addition, consumers generally value organic, larger size (as long as they are not too large),

designated fancy grade and a heart healthy claim for pecan purchases. However, these results

show significant taste heterogeneity across consumers for many pecan attributes, which must be

considered in anticipation of consumer responses to any proposed initiatives. Findings indicate

that, in choice experiments, consumer ethnocentric tendencies and purchasing patterns play a

role in defining differences among consumer taste preferences and WTP, while efforts to inform

and educate consumers about attributes are essential.

In Chapter 2, we examine how much consumers are willing to pay for knowing vs. not

knowing information on the label for pecan purchases using a direct approach in an experimental

setting. The direct value of information (VOI) method measures how much consumers will pay

to have additional information about their product purchase rather than the typical approach

measuring consumer valuation of specific attributes. On average, we find that consumers have a

positive VOI for country of origin information, even without knowing exactly which country, yet

the median individual has none. How consumer VOI varies in the presence of other knowledge

94

about the product is not clear from the available data, though there is some evidence that the

value of origin information diminishes (or at least changes) in a more complex label

environment.

In Chapter 3, we employ a series of experimental auctions to elicit the consumer

willingness to pay for pecans under labeling scenarios that include single and mixed country of

origin products, both with and without the risk and/or ambiguity of knowing the country of

origin. We found that, on average, consumers value most the single origin product from the U.S.

with certainty over any other option and least value the single origin product from Mexico. In

comparing mixed country of origin options, consumers are indifferent between two mixed

product label options and in the presence of risk or ambiguity regarding knowledge of product

provenance, consumers prefer mixed origin over any risk of obtaining the single origin product

from their lowest ranked preference country.

95

APPENDIX A: PARTICIPANT INFORMATION PACKET

Welcome to our research session about decision making. Thank you for choosing to participate. The information you provide today is a very important contribution to ongoing research by the University of Georgia. Enclosed is the packet of information that you will need during the session. Once you have looked at a form during the session, feel free to go back and reexamine the form again if needed, but please do not look ahead until we reach the right point in the session. Please follow instructions in this packet carefully. To ensure accuracy, we request that you do not talk to any other participants during the session. We would like to emphasize that all information obtained today will be used only for group comparisons. No personal or individual information will be divulged for any reason. Please turn to the next page.

96

Please answer the following questions by circling the appropriate choice or filling in the appropriate line.

1. What is your gender? 1 = Male 2 = Female

2. What is your age? ______ 3. What best describes your marital status?

1 = Single 2 = Married 3 = Divorced, widowed, or separated 0 = Other

4. How many people live in your household (including yourself)? ______ 5. What is the highest level of schooling that you have completed?

1 = Some high school 2 = Graduated from high school/GED 3 = Some college 4 = 2 year college degree 5 = 4 year college degree 6 = Master’s Degree 7= Doctoral Degree 8= Professional Degree (JD, MD)

6. What is your racial-ethnic background?

1 = Native American or Alaska Native 2 = White or Caucasian (non-Hispanic) 3 = African-American 4 = Asian-American 5 = Hispanic or Latino 6= Native Hawaiian or Pacific Islander 0 = Other (please fill in) ____________________

7. What was your total household income (before taxes) in 2015?

1 = Under $24,999 2 = $25,000-$49,999 3 = $50,000-$99,999 4 = $100,000-$149,999 5 = $150,000 or more

Please do not turn the page until instructed by your monitor.

97

Once again, we would like to thank you for participating in today’s session about decision making. We will be conducting auctions of some common products. Details for how the auction works will be provided shortly. Because we are trying to determine values for products, we ask that you please refrain from communicating with the other participants. If you have any questions, the monitors can assist you.

98

How the Auction Works

We are going to hold what is called an nth price auction. For those of you who have participated in auctions before, please note that the nth price auction is slightly different than what you may have previously encountered. The nth price auction works as follows: 1. We will pass out bid sheets explaining information about products on which to

bid. Carefully examine the bid sheet, ensuring that it contains the ID# you were assigned.

2. Write down your bid for each product

After examining the bid sheet, write down what dollar amount (if any) you would like to bid for the products being auctioned on the provided bid sheet.

3. Choosing of the nth price

Once everyone has bid, we will collect the bids and determine what will be called the “nth price”. Everyone who bids higher than this price will win the product, and pay the nth price. (Your monitor will go through an example of this)

4. Determining who wins the auction

(Your monitor will go through an example of this)

Please note that in this auction it is always in your best interest to bid your true value for a product. Unlike many auctions in which you might bid less than your true value to try to get a deal, this auction does not reward that strategy. This is because you do not necessarily pay your bid price, but you pay the nth price that is randomly chosen. Likewise, it is not in your interest to bid more than you are truly willing to pay because you may have to pay more than you wanted to for the product.

99

Short Quiz on the nth Price Auction Format

Please note: this sheet will remain in your packet

1. The people who win will always pay the amount they bid for a product. 1 = True 2 = False

2. If you have the 4th highest bid and the randomly drawn nth price is the 2nd

highest, you will win the auction. 1 = True 2 = False

3. I might get to pay less than my bid for a product, but I will never have to pay

more than my bid for a product. 1 = True 2 = False

4. If the binding price that is randomly drawn is the 7th highest price, how many

people win the good? 4 5 6 7 8

100

Practice Auction To make sure everyone is comfortable with how the nth price auction works we will have two rounds of practice bidding. Since some of the products in the two rounds are similar, only one of the two practice rounds will be “binding”. By “binding” we mean that only one of the two practice rounds will be selected as the round where people will win goods and pay money for them (i.e. only one round “counts”). The round that is binding will be randomly selected and will be revealed after the second practice round. Since you do not know which round will be chosen, it is in your best interest to bid your true value for the products in both practice rounds. These two rounds are practice so no goods will actually be purchased and no money will be exchanged. Practice Bidding Round 1 of 2 Step 1 - Examine the 2 products Examine the products in practice round 1 Step 2 - Write down your bid Please fill out your bid sheet. It will be collected by monitors.


101

Practice Bidding Round 2 of 2

Step 1 - Examine the 2 products Examine the products in practice round 2 Step 2 - Write down your bid Please fill out your bid sheet. It will be collected by monitors. Step 3 – Determine the binding round (randomly selected)

Monitors will randomly determine which of the two rounds of bidding will be binding.

Step 4 – Determine the nth price for each product (computer generated)

Monitors will randomly determine the nth price for products in the binding round.

Step 5 – Announcement of the auction winners for each product

If this auction was real, the winners in the binding round would exchange money for the goods in the room next door.


102

Auction We are about to begin the real auction. The auction will consist of 6 products for which you will have the opportunity to place a bid. However, only one set of product bids will be “binding” (i.e., winners of the auction will pay for their products). The binding product will be determined randomly after all bidding is completed. You will compare 8 oz. bags of pecan halves that are all of the same weight and consistency. However, each of the 6 products will contain different labeling information than the other. You will bid the highest price you are willing to pay for each bag of pecans given the characteristics. At the end one product will be selected randomly as binding. If you are a winner of the auction, then you will receive pecans with the randomly selected characteristics and pay your bid as demonstrated in the practice round. Remember, please bid with your true value.

103

Bidding Procedure Step 1 - Examine the information/bid sheet for 6 products

Examine the products, reading carefully the information explaining the differences between the products

Step 2 - Write down your bid for each product

Please fill out your bid sheet. It will be collected by monitors. To ensure accuracy, write your bids here also:

Option A Option B The highest price that I am willing to pay is The highest price that I am willing to pay is:

$_________________ $__________________

Option C Option D The highest price that I am willing to pay is The highest price that I am willing to pay is:

$_________________ $__________________

Option E Option F The highest price that I am willing to pay is The highest price that I am willing to pay is:

$_________________ $__________________


104

Step 3 – Determine the binding product (randomly selected)

Monitors will randomly determine which of the product bids will be binding

Step 4 – Determine the nth price for each product (randomly selected)

Researchers will randomly determine the nth price for the binding product bids

Step 5 – Announcement of the auction winners Step 6 – Post auction questionnaire

Please fill out the questionnaire on the following pages. Once you have completed the questionnaire, please return your information packet to the session monitor.

Step 7 – Auction winners exchange money for goods


105

The following questions represent different descriptions for the purchase of an 8 oz. bag of pecan halves. Please check the option which you would be most likely to purchase. C- 1.1:

C- 1.2:

C- 1.3:

Option A Option B Option C Price $2.25 $5.00

Neither A nor B is preferred

Organic? Yes No Expiration Date 6 Months 3 Months Country of Origin U.S.A. Mexico Size Jumbo Halves Small Halves Grade Choice Standard Health/Nutrient Claim Heart Healthy I would choose…



Organic? No Yes Expiration Date 6 Months 6 Months Country of Origin U.S.A. Mexico Size Large Halves Large Halves Grade Fancy Health/Nutrient Claim Heart Healthy I would choose…



Organic? Yes No Expiration Date 3 Months 12 Months Country of Origin Mexico U.S.A. Size Small Halves Extra Large Halves Grade Fancy

Health/Nutrient Claim Naturally High in Antioxidants

Naturally High in Antioxidants

I would choose…

106


C-1.5:

C-1.6:



Organic? No Yes Expiration Date 12 Months 3 Months Country of Origin U.S.A. Mexico Size Mammoth Halves Grade Fancy



I would choose…



Organic? No Yes Expiration Date 6 Months 6 Months Country of Origin Mexico U.S.A. Size Extra Large Halves Small Halves Grade Standard Choice



I would choose…



Organic? Yes No Expiration Date 12 Months 3 Months Country of Origin Mexico U.S.A. Size Mammoth Halves Grade Fancy Health/Nutrient Claim Heart Healthy I would choose…

107


C- 1.8:



Organic? No Yes Expiration Date 3 Months 12 Months Country of Origin Mexico U.S.A. Size Extra Large Halves Small Halves Grade Choice Standard Health/Nutrient Claim Heart Healthy I would choose…



Organic? Yes No Expiration Date 3 Months 12 Months Country of Origin U.S.A. Mexico Size Small Halves Jumbo Halves Grade Standard Choice



I would choose…

108

Please answer the following questions by circling the appropriate choice or filling in the appropriate line.

8. Are you aware of mandatory country of origin labeling for pecan products? 1 = Yes 2 = No 3 = I don’t know

Rate the following three statements on a scale of 1 to 5, with 1 being strongly disagree and 5 being strongly agree.

9. Americans should not buy foreign products, because this hurts American business and causes

unemployment. 1=Strongly Disagree 2=Disagree 3=Neutral 4=Agree 5=Strongly Agree

10. It is not right to purchase foreign products because it puts Americans out of jobs. 1=Strongly Disagree 2=Disagree 3=Neutral 4=Agree 5=Strongly Agree

11. A real American should always buy American made products. 1=Strongly Disagree 2=Disagree 3=Neutral 4=Agree 5=Strongly Agree

12. Do you read the nutrition facts label printed on the food packages you consume? 1 = Always 2 = Often 3 = Sometimes 4 = Rarely 5 = Never

13. Do you look for expiration dates printed on the food packages you consume? 1 = Always 2 = Often 3 = Sometimes 4 = Rarely 5 = Never

109

14. Do you look for country of origin printed on the food packages you consume?

1 = Always 2 = Often 3 = Sometimes 4 = Rarely 5 = Never

15. On a scale of 1 to 10, with 1 being unhealthy and 10 being very healthy, how healthy is your

diet? ______ 16. On a scale of 1 to 10, with 1 being unhealthy and 10 being very healthy, how do you consider

your physical health? ______ 17. How do you see yourself: are you generally a person who is fully prepared to take risks or do

you try to avoid taking risks? Please check a box on the scale, where the value 0 means: ‘not at all willing to take risks’ and the value 10 means: ‘very willing to take risks’.

0 1 2 3 4 5 6 7 8 9 10

18. On average, how often do you purchase organic foods? 1=Very Often 2=Often 3=Sometimes 4=Rarely 5=Very Rarely

19. On average, how often do you consume pecans? 1=Very Often 2=Often 3=Sometimes 4=Rarely 5=Very Rarely

20. Select which statement best describes the form in which you consume pecans.(circle all that apply) 1=Pecans as a snack by themselves 2=Pecans along with other nut products (mixed nuts) 3=Pecans as ingredients in cooking and baking 4=Pecans in prepared meals from restaurants 5=Pecans in other forms

21. Please circle the products that you currently have in your home. (circle all that apply) 1 = Pecans as a snack by themselves 2 = Pecans along with other nut products (mixed nuts) 3 = Pecans as ingredients for cooking and baking 4 = Pecans in prepared forms such as restaurant meals or baked goods

110

5 = Other forms

22. What price would you expect to pay for a ½ lb. package of shelled pecan halves at the store? $_______

23. On a scale of 1 to 10, where 1 means “very uncertain” and 10 means “very certain,” how certain are you that the price you listed above is within 50¢ of the price of pecans at the grocery store? _______

24. On a scale of 1 to 10, where 1 means “very uncertain” and 10 means “very certain,” how

certain are you that all of your auction bids today were equal to your true maximum willingness to pay for the different types of pecans? _______


111

Instructions for Task A In addition to these Instructions, you are provided with a Decision Sheet. Please look over the

Decision Sheet as you read these Instructions to ensure that you understand the procedures involved in this task Let the monitor(s) know if you have a question by raising your hand.

The Decision Sheet contains 10 separate Decisions numbered 1 through 10. Each of these Decisions is a choice between “Option A” and “Option B”. One of these decisions will be randomly selected to determine how much money you will receive. In order to select one of the choices, a ten-sided die will be used to determine the payoffs. After you have made your choice, this die will be rolled twice: once to select one of the 10 Decisions to be used, and then again to determine your payoff for the Option associated with that decision, either A or B, given your choice at that decision.

To choose an Option for each decision, you will make one selection in the “Choice” column on the right. This choice indicates whether you would choose Option A or Option B, and will signify whether Option A or Option B will be used to determine your earnings for each of the 10 decisions.

For example, if the die roll outcome is 6, Decision No. 6 would determine payment.

1. If your “Choice” for No. 6 is A, then Option A would be used to determine your payoff. You would have a 6/10 chance of earning $2.00, and a 4/10 chance of earning $1.60.

2. If your “Choice” for No. 6 is B, then Option B would be used to determine your payoff. You would have a 6/10 chance of earning $3.85, and a 4/10 chance of earning 10¢.

As an example, look at No. 3 on the Decision Sheet. You can see that Option A pays $2.00

with a chance of 3/10, and $1.60 with a chance of 7/10. Since each side of a ten-sided die has an equal chance of being the outcome in a throw, this corresponds to Option A paying $2.00 if the throw of the die is 1, 2 or 3, and $1.60 if the throw of the die is any other number (4 through 10). Option B pays $3.85 if the throw of the die is 1, 2 or 3, and 10¢ if the throw of the die is any other number (4 through 10). All of the other choices are similar, except that as you go down the table, the chances of the higher payoff for each Option increase. For Decision 10 in the bottom row, no die will be needed since each Option pays the highest payoff for sure. Your choice there is between $2.00 and $3.85.

Once you are done with both tasks A and B, you will proceed to another room where an

experimenter will flip a coin. If the outcome is Heads, the experimenter will throw a ten-sided die to select which of the ten Decisions will be used. The die will then be thrown again to determine your earnings for the Option you chose for the selected Decision. You will be paid in cash when finished.

Please turn over these instructions so that the experimenter knows that you have finished reading

them. If you do not have any questions, please proceed to the Decision Sheet and mark your choices.

112

Task A Decision Sheet

This is your Decision Sheet. Please indicate at each decision number whether you would like to choose Option A or Option B by putting a check mark in the box of the Choice column. When you are finished, you should have only 1check mark for each row in the Choice column.

No. Option A Option B Choice

1 1/10 chance of $2.00 9/10 chance of $1.60

1/10 chance of $3.85 9/10 of 10¢

Option A Option B


2/10 chance of $3.85 8/10 of 10¢

Option A Option B


3/10 chance of $3.85 7/10 of 10¢

Option A Option B


4/10 chance of $3.85 6/10 of 10¢

Option A Option B


5/10 chance of $3.85 5/10 of 10¢

Option A Option B


6/10 chance of $3.85 4/10 of 10¢

Option A Option B


7/10 chance of $3.85 3/10 of 10¢

Option A Option B


8/10 chance of $3.85 2/10 of 10¢

Option A Option B


9/10 chance of $3.85 1/10 of 10¢

Option A Option B


10/10 chance of $3.85 0/10 of 10¢

Option A Option B

113

Instructions for Task B

In addition to the Instructions, you are also provided with a Decision Sheet. Please look over

your Decision Sheet as you read these Instructions to ensure that you understand the procedure of the experiment. If you have a question at any point, please raise your hand.

The Decision Sheet contains 20 separate Decisions that are numbered from 1 through 20. Each of these Decisions is a choice between drawing a ball from “Urn A” or “Urn B”. One of these decisions would be randomly selected to determine how much money you would receive. You will first select a color, Blue or Green, and this will be your Success Color. Once a decision is selected, your earnings will be determined by whether the ball drawn from the Urn matches your Success Color.

After choosing your Success Color, you will make one choice for each row in the “Choice” column on the right. This choice indicates whether you would like to choose between drawing a ball out of Urn A or drawing out of Urn B.

After choosing your Success Color, you will make one choice for each row in the “Choice” column on the right. This choice indicates whether you would like to choose between drawing a ball out of Urn A or drawing out of Urn B.

For example, if the dice roll outcome is 9, Decision No. 9 would determine payment. 1. If your “Choice” is Urn A, a ball would be drawn from Urn A, and if the color of the

ball matches the chosen Success Color, then you would earn $2.00. If it does not match, you would earn 0.

2. If your “Choice” is Urn B, a ball would be drawn from Urn B, and if the color of the ball matches the chosen Success Color, then you would earn $2.28. If it does not match, you would earn 0.

In each of the 20 decisions, Urn A has 50 Blue balls and 50 Green balls, and pays $2.00 if the

ball drawn from Urn A matches your Success Color, and 0 if it does not match. Since each color has a 1/2 chance of being drawn, this means that drawing from Urn A pays $2.00 with a chance of 1/2, and pays 0 with a chance of 1/2.

Urn B, on the other hand, has an unknown number of Blue and Green balls (with a total of 100 balls). It pays a positive payout amount if the ball drawn from Urn B matches your Success Color, and 0 if it does not match. Since the chance of each color being drawn is unknown, the chance of Urn B paying a positive payout amount is unknown as well. The only difference between the 20 options is the amount paid when a ball matching your Success Color is drawn from Urn B.

Once you are done with both tasks A and B, you will proceed to another room where an

experimenter will flip a coin. If the outcome is Tails, the experimenter will throw one twenty-sided dice to select which of the 20 decisions will be used. The experimenter will then draw a ball from the Urn you had selected for that Decision to determine the payoff. You will then be paid in cash.

When you have finished reading the Instructions and do not have any questions, please proceed

to the Decision Sheet and mark your choices.

114

Task B Decision Sheet

My Success Color is (please circle one): Blue Green Please indicate at which decision you would like to switch from Urn A to Urn B by putting a check mark in the box of the Switch column. When you are finished, you should have only 1check mark in the Switch column. For any decisions before this check mark, a ball will be drawn from Urn A. For any decisions after and including the check mark, a ball will be drawn from Urn B.

Urn A Urn B Choice

No. 50 Blue balls, 50 Green balls ? Blue balls, ? Green balls (select 1 per row)

1 $2.00 if Chosen Color 0 if not

$1.64 if Chosen Color 0 if not

Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B


$2.04 if Chosen Color 0 if not 60

Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B



Urn A Urn B

115

Thank you for your participation! Please return your entire packet to the monitors

Date post:	10-Dec-2021
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

THE INFLUENCE OF INFORMATION AND PRODUCT LABELS ON CONSUMER

Documents